1. AI驱动的药物研发与精准医疗创新
1.1 药物发现与开发的AI赋能
人工智能（AI）和机器学习（ML）技术正在深刻变革药物发现与开发的全流程，通过提升效率、降低成本并缩短研发周期。传统药物研发面临着效率低下、成本高昂、脱靶效应以及耗时漫长等挑战，而AI的应用为解决这些问题提供了强大的工具 12。
在药物发现的早期阶段，AI主要赋能于虚拟筛选和分子设计。虚拟筛选利用AI算法从海量的化合物库中快速识别潜在的活性分子。例如，基于深度学习的生成模型（如深度生成模型, DGMs）在全新药物设计（de novo drug design）中展现出巨大潜力，能够生成具有特定药理活性的新型分子结构 3。这包括预测分子与靶点的结合亲和力，以及优化分子的理化性质 4。多项研究表明，AI技术已被广泛应用于肽合成、基于结构和基于配体的虚拟筛选、药效团建模以及定量构效关系（QSAR）分析等环节 15。
此外，AI在药物的毒性预测方面也发挥着关键作用。药物毒性是导致药物研发失败的主要原因之一，约30%的候选药物因毒性问题被淘汰 67。AI模型能够通过分析大规模毒理学、化学和多组学数据，预测化合物的潜在毒性，从而在早期阶段筛选掉不安全的候选药物 6。例如，MolToxPred工具利用机器学习方法预测小分子和代谢物的潜在毒性，在测试集上取得了87.76%的AUROC 8。图像识别技术结合卷积神经网络（CNN）也被应用于肝毒性预测，利用三维构象图像进行分析，提高了预测准确性 9。这些AI驱动的毒性预测方法不仅能节约实验成本和时间，还能减少动物实验，符合伦理要求 8。
天然产物作为新药的重要来源，其药物发现过程也受益于AI技术。计算组学技术的发展使得天然产物隐藏的多样性得以被挖掘，而AI，特别是机器学习，则促进了生物活性预测和全新药物设计 1011。DerivaPredict工具能够自动生成天然产物衍生物，并通过预训练的深度学习模型预测结合亲和力，评估药物类似性，极大地加速了天然产物药物的早期发现 12。
在传统中药（TCM）领域，网络药理学（NP）与AI的结合正在推动中药现代化。中药以其“多成分、多靶点、多通路”的特点治疗复杂疾病，AI驱动的网络药理学方法能够从整体层面揭示中药的作用机制 1314。AI在网络关系挖掘、靶点定位和网络靶点导航中发挥作用，例如，FordNet系统融合表型和分子信息，利用深度学习模型推荐中药方剂，其推荐准确率远超传统基线方法，并能有效学习中医名家的临床经验 15。最近的研究还展示了机器学习增强的网络药理学工作流，通过超高效液相色谱-四极杆飞行时间质谱技术识别中药化学成分，并利用机器学习模型预测具有抗过敏性鼻炎活性的化合物，从而系统性地筛选关键成分进行网络分析，为中药作用机制的阐明提供了更精准的手段 16。
1.2 临床试验与个性化用药优化
人工智能（AI）正在革新临床试验的设计、患者招募、疗效预测以及个性化用药策略，从而显著提升试验效率和用药精准性。传统临床试验面临耗时久、成本高、成功率低以及患者招募困难等挑战，而AI的应用为克服这些难题提供了新的途径 1718。
在临床试验设计方面，AI能够优化试验方案，包括确定最佳剂量、治疗持续时间以及评估指标。通过模拟和预测不同设计方案的潜在结果，AI可以帮助研究人员设计出更有效、更具成本效益的试验。例如，自适应设计和主方案（如伞式、篮式和平台试验）等创新试验设计正逐步取代传统的I、II、III期方案，它们结合了AI和ML技术，能够加速药物评估和审批过程，并实现基于分子特征的个体化治疗 1819。
患者分层是实现精准医疗的关键一步，AI在此领域展现出强大能力。通过分析患者的基因组、转录组、蛋白质组、免疫组等多组学数据以及临床特征，AI模型可以识别出对特定治疗反应更佳的患者亚群 1920。这有助于将合适的药物匹配给合适的患者，提高治疗成功率并减少不必要的副作用。例如，在一项预测精神分裂症患者对帕利哌酮治疗反应的研究中，可解释AI（Explainable AI, XAI）方法结合“个性化优势指数”和“贝叶斯规则列表”识别出对药物反应更佳的患者亚群，显著提升了治疗效果（Cohen's d从全样本的0.82增至亚组的1.22）21。该模型甚至能生成简单的“if-then-else”规则，提高了临床医生对预测结果的理解和信任度 21。另一个案例是，通过对鼻病毒刺激的个体化转录组反应进行分类，AI模型成功预测了儿童哮喘恶化的风险，准确率达到74%，这表明“基因-环境”交互作用在表型呈现中的重要性，并为精准医疗提供了创新方法 22。
在疗效预测方面，AI利用复杂的算法分析患者的基线数据和治疗期间的生物标志物变化，预测治疗效果 23。这不仅有助于优化治疗方案，还能为临床医生提供决策支持。例如，机器学习已被用于预测心力衰竭患者对心脏再同步治疗（CRT）的反应，通过整合人口统计学、生物标志物和左心室功能等多维度数据，模型能够以71%的准确率预测CRT反应，并能以96%的准确率识别出最有可能或最不可能受益的患者亚组 24。此外，AI和ML在治疗药物监测（TDM）和模型引导的精准给药（MIPD）中也发挥着关键作用，例如通过预测药物浓度、优化剂量以及辅助群体药代动力学和药效学研究等 25。强化学习作为一种AI技术，已被用于危及生命疾病患者的个体化给药方案，通过优化给药策略以实现最佳治疗效果 26。
真实世界证据（RWE）分析结合AI，正在改变药物评估和监管决策的模式。RWE来源于日常临床实践中收集的真实世界数据（RWD），如电子健康记录（EHR）、医保理赔数据、患者报告结果（PROs）以及可穿戴设备数据等 272829。虽然随机对照试验（RCTs）仍是评估药物疗效的金标准，但RWE能够补充RCTs的不足，提供更广泛患者群体和更长期的治疗效果信息 193031。AI，特别是自然语言处理（NLP）和大型语言模型（LLMs），可以自动化RWD的收集、提取、清洗和分析过程，克服数据不一致性和质量问题，从而释放RWE的全部潜力，加速新药研发和上市 28。例如，在肿瘤学领域，AI技术有望通过提升RWE的质量和可信度，加速罕见癌症的药物开发，并为更稳健的肿瘤学研究提供支持 28。
临床药理学数据建模结合AI，能够更深入地理解药物作用机制和个体差异 2032。通过建立复杂的药代动力学/药效学（PK/PD）模型，AI可以模拟药物在体内的吸收、分布、代谢和排泄过程，并预测其药理效应。这有助于识别药物反应的生物标志物，优化给药方案，并为精准医疗提供科学依据 232532。然而，利用电子健康记录（EHR）进行临床药理学研究仍面临挑战，尤其是在药物相关表型数据的准确提取和标准化方面，这需要计算表型和自然语言处理等前沿方法的应用 33。
2. 蛋白质与基因组工程的AI突破
2.1 功能蛋白质的AI设计与定向进化
人工智能（AI）的最新突破，特别是深度学习和蛋白质语言模型（PLMs），正在彻底改变功能蛋白质的设计与定向进化领域，加速了生物制造和医学应用中酶、抗体等功能蛋白的开发进程 3435。传统的蛋白质工程方法往往效率低下、耗时且受限于局部最优解，而AI技术通过对蛋白质序列、结构和功能之间复杂关系的建模，显著提升了设计效率和成功率 36。
AI驱动的蛋白质设计通常涉及以下核心环节：
序列-结构-功能关联建模：AI模型能够学习蛋白质序列与其三维结构以及生物功能之间的内在联系。例如，通过分析大规模蛋白质序列和结构数据集，深度学习模型可以预测给定序列的蛋白质结构，反之亦然，甚至能够从头设计（de novo design）全新的蛋白质结构 37。RoseTTAFold等深度学习方法已被用于“约束幻觉”（constrained hallucination）和“填补”（inpainting）技术，在不预设骨架或二级结构的情况下，设计包含特定功能位点的蛋白质支架，从而成功设计出免疫原、受体陷阱、金属蛋白、酶和蛋白质结合蛋白等 38。这种能力使得研究人员能够设计出自然界中不存在的，但具有特定所需功能的蛋白质。
多模态数据整合：蛋白质设计不再局限于单一数据类型，AI模型能够整合包括序列信息、结构信息（如AlphaFold预测结构）、物理化学性质、以及实验表征数据（如活性、稳定性、溶解度）等在内的多模态数据，以实现更全面的蛋白质功能预测和优化 3439。例如，DDGemb方法结合蛋白质语言模型嵌入（PLM embeddings）和Transformer架构，能够预测单点和多点突变对蛋白质稳定性的影响，为功能蛋白质设计提供了关键的稳定性评估工具 40。
酶的设计与优化：AI在酶工程中表现出巨大潜力，旨在提升酶的催化活性、稳定性、底物特异性以及在特定工业条件下的耐受性。例如，FAST-PETase是一种通过机器学习算法设计的PET水解酶，其在30-50°C温度范围和不同pH条件下展现出优于野生型和现有工程酶的PET水解活性，能在一周内几乎完全降解51种不同的消费后PET塑料，并实现了PET的闭环回收 41。ProT-VAE模型利用Transformer变分自编码器成功设计出一种苯丙氨酸羟化酶，其催化活性是野生型的2.5倍，以及一种碳酸酐酶，其熔点温度提升了61°C，并在工业相关条件下（如93°C，pH 11.25，23% (v/v) 甲基二乙醇胺）仍保持活性 42。这些案例展示了AI在加速工业级生物催化剂应用方面的巨大潜力。
抗体的设计与亲和力优化：抗体作为一类重要的治疗性蛋白，其设计与优化是生物制药领域的关键。AI能够加速抗体发现过程，通过预测抗原特异性、优化亲和力、降低免疫原性以及提高稳定性。AbDPP（target-oriented antibody design with pretraining and prior biological structure knowledge）是一种新型的深度学习方法，它整合了预训练的抗体模型和生物区域信息，能够有效利用大量抗体序列数据，生成具有特定抗原结合特性的抗体序列，并在性能和生成序列质量上超越了其他方法 4344。通过深度学习，研究人员可以从大规模抗体库中识别出具有高特异性的抗体变体，显著加速抗体工程和优化过程 454647。
定向进化框架：AI与定向进化相结合，形成了强大的循环优化策略。EVOLVEpro是一个少样本主动学习框架，它结合了蛋白质语言模型和回归模型，能够快速提升蛋白质活性 3648。该框架只需少量实验数据，就能实现蛋白质目标性质（如RNA生产、基因编辑、抗体结合）高达100倍的改进，展示了AI在加速蛋白质工程中的巨大优势。另一个名为“松弛序列空间优化”（optimization in a relaxed sequence space）的设计方法，通过扩散生成模型，无需重新训练即可高效设计高质量的蛋白质骨架，并已成功设计出单链蛋白和蛋白质异二聚体，这些设计的准确性已通过高分辨率晶体结构和冷冻电镜密度图验证 49。
神经蛋白设计：AI甚至被用于设计具有计算功能的人工蛋白质网络。例如，研究人员通过设计蛋白质异二聚体和工程化病毒蛋白酶，构建了一种在哺乳动物细胞中实现“赢者通吃”神经网络分类的合成蛋白质回路。这种“perceptein”回路能够通过可逆结合和不可逆蛋白水解，实现多输出信号分类，并可用于条件性控制细胞死亡 50。这预示着AI在设计复杂生物系统和合成生物学方面的广阔前景。
这些AI驱动的功能蛋白质设计与定向进化技术，正为生物制造（如新型生物催化剂、生物材料）、医疗（如新型抗体药物、疫苗、基因治疗工具）等领域带来前所未有的创新机遇。
2.2 基因组编辑效率与安全性的AI优化
基因组编辑技术，特别是CRISPR-Cas系统，已成为生命科学和生物技术领域的革命性工具，为疾病治疗和作物改良带来了前所未有的可能性。然而，提高编辑效率、确保编辑特异性（即减少脱靶效应）以及优化递送方式仍是该技术广泛应用的关键挑战。人工智能（AI）和机器学习（ML）的引入，正在为解决这些问题提供强大的计算解决方案，从而显著提升基因组编辑的效率、安全性和精准度 515253。
pegRNA设计与编辑结果预测的AI优化
Prime Editing (PE) 是一种先进的基因编辑技术，能够实现精准的碱基替换、小片段插入和删除，而无需DNA双链断裂。PE的成功高度依赖于引物编辑向导RNA（pegRNA）的设计和目标基因座的特性。为了优化pegRNA的设计并预测其编辑效率，研究人员开发了多种基于AI的预测模型：
- PRIDICT模型：该模型利用深度学习，特别是注意力机制的双向循环神经网络，通过高通量筛选分析了92,423个pegRNA在13,349个人类致病突变上的编辑结果。PRIDICT能够可靠地预测所有小尺寸基因变化的编辑效率（预期编辑R值为0.85，非预期编辑R值为0.78），并且在体内外实验中都得到了验证，显示出高PRIDICT评分的pegRNA能显著提高Prime Editing效率（体外12倍，体内10倍）54。
- PRIDICT2.0和ePRIDICT：这是PRIDICT的升级版本，进一步评估了在错配修复缺陷和错配修复正常细胞系以及原代细胞中，长度达15 bp的所有编辑类型的pegRNA性能。ePRIDICT模型还量化了局部染色质环境如何影响Prime Editing的效率，为更精准的pegRNA设计提供了依据 55。
除了Prime Editing，AI也被广泛应用于CRISPR-Cas9等其他基因编辑系统的向导RNA (gRNA) 设计和编辑效率预测：
- DeepCRISTL：这是一个基于深度迁移学习的模型，用于预测CRISPR/Cas9在靶点上的功能性和内源性编辑效率。它通过在大规模高通量数据集上进行预训练，然后利用迁移学习在较小的、更具生物学相关性的功能性和内源性数据集上进行微调，从而显著提高了预测性能 56。
- BoostMEC：该模型采用梯度提升方法，利用LightGBM算法预测野生型CRISPR-Cas9的编辑效率。它在多个外部数据集上表现出与现有深度学习模型相媲美的性能，同时通过可解释的特征洞察力，帮助理解不同特征对CRISPR-Cas9切割效率的贡献 57。
- CRISPRon：通过整合高质量的gRNA活性数据（包括10,592个SpCas9 gRNA的新数据）和深度学习模型，CRISPRon在多个测试数据集上展现出显著更高的预测性能，为更准确的gRNA效率预测提供了工具 58。
脱靶风险评估的AI应用
脱靶效应是基因编辑技术面临的主要安全挑战之一，可能导致非预期的基因突变。AI在预测和最小化脱靶效应方面发挥着关键作用：
- TIGER (Targeted Inhibition of Gene Expression via gRNA Design)：该卷积神经网络模型通过分析约20万个RfxCas13d gRNA在人类细胞中的靶向数据，揭示了错配和插入/缺失对Cas13d活性的位置和上下文依赖性影响。TIGER在预测Cas13d的在靶和脱靶活性方面优于现有模型，并为调节转录表达提供了通用框架 59。
- CCLMoff：这是一个利用预训练RNA语言模型（来自RNAcentral）的深度学习框架，用于预测CRISPR/Cas9系统的脱靶效应。CCLMoff能够捕获sgRNA和靶点序列之间的相互信息，并在各种基于NGS的检测数据集上实现了准确的脱靶识别和强大的泛化能力。模型解释还揭示了种子区域的生物学重要性 60。
- R-CRISPR：这是一个深度学习网络，能够预测CRISPR-Cas9系统在存在错配、插入和删除情况下的脱靶活性。该模型通过特殊的编码方案和卷积神经网络作为特征提取器，在预测脱靶效应方面显著优于现有方法 61。
- ABEdeepoff和CBEdeepoff：针对碱基编辑器（Base Editors, BEs），这些深度学习模型通过分析大规模gRNA-脱靶对数据，预测腺嘌呤碱基编辑器（ABEs）和胞嘧啶碱基编辑器（CBEs）的脱靶位点，并实现了较高的斯皮尔曼相关系数（0.710至0.859） 62。
- DeepCas13：用于预测CRISPR-Cas13d在靶和脱靶效应的深度学习模型，通过分析Cas13d增殖筛选数据，能够准确预测gRNA序列和二级结构对在靶活性的影响，并识别出影响细胞活力和增殖的长非编码RNA 63。
- AI驱动的表观遗传CRISPR疗法：AI显著提高了基于CRISPR的表观遗传编辑工具的精度、安全性和治疗效果。一项针对2015-2025年研究的荟萃分析显示，AI在治疗效果、gRNA优化和脱靶预测方面均表现出强大的积极影响，特别是深度学习模型与更高的效应值相关联 64。
在基因治疗和作物育种中的技术突破
AI赋能的基因组编辑优化不仅推动了基础研究，也为实际应用带来了变革：
- 基因治疗：AI通过优化CRISPR-Cas系统的设计，有助于开发更安全、更有效的基因疗法。例如，AI驱动的衣壳设计可以提高病毒载体的靶向特异性并降低免疫原性，从而优化基因递送效率，这对于治疗遗传性疾病至关重要 6566。AI还被用于选择高效的Cas9变体和gRNA，以实现高保真和高效的基因编辑，并通过机器学习模型预测gRNA在人类基因组位点中的编辑效率 67。
- 作物育种：CRISPR技术已在水稻、小麦、玉米和大麦等主要谷物作物中取得显著进展，用于增强各种农艺性状。AI与CRISPR的结合进一步加速了这一进程，例如，AI工具被用于优化编辑系统、预测编辑位点效率和设计编辑策略，从而培育出抗病、高产、适应性更强的作物，为全球粮食安全和可持续农业做出了贡献 5368。AI和ML技术还被用于识别基因型和表型数据中的新序列变异，以加速植物育种过程 69。
综上所述，AI在基因组编辑领域，尤其是CRISPR技术中的应用，正从gRNA设计、编辑效率预测到脱靶风险评估，全方位提升了该技术的精准性、安全性和应用潜力，为基因治疗和农业创新带来了广阔前景。
3. 免疫与疾病防控的AI融合研究
3.1 免疫学研究的AI范式转变
免疫学是理解疾病发生发展、开发新型疗法和疫苗的关键领域。近年来，人工智能（AI）的融合正在为免疫学研究带来范式转变，特别是在处理和解析高维、多尺度免疫数据方面展现出无与伦比的优势。AI不仅加速了对免疫系统复杂性的理解，也推动了从基础研究到临床应用的转化，尤其是在免疫数据挖掘、疫苗设计以及个性化肿瘤免疫治疗等方面 70。
AI在免疫数据挖掘中的应用
免疫系统的数据复杂多样，涵盖了基因组、转录组、蛋白质组、细胞因子谱以及T细胞受体（TCR）/B细胞受体（BCR）库等多个层面 70。AI技术能够高效地整合和分析这些异构数据，从中发现隐藏的模式和生物学见解：
- T细胞受体（TCR）库分析：TCR是T细胞识别抗原的关键分子，其多样性反映了免疫系统对病原体和肿瘤的识别能力。AI，特别是机器学习算法，能够对海量的TCR序列数据进行聚类、分类和特异性预测。例如，GLIPH（grouping of lymphocyte interactions by paratope hotspots）算法能够根据TCR序列的保守基序和CDR3序列的全局相似性，将具有共同抗原特异性的TCR进行分组，即使是来自不同个体的TCR也能被有效识别 71。GLIPH2模型进一步扩展了这一能力，通过分析178名非小细胞肺癌患者的778,938条TCRβ链序列，识别出超过66,000个共享特异性组，并成功鉴定了肿瘤相关抗原表位，揭示了病原体交叉反应性在多种癌症中的潜在作用 72。这些AI驱动的分析极大地加速了TCR抗原特异性的识别，有助于理解T细胞在疾病中的作用。
- 免疫检查点机制研究：免疫检查点抑制剂（ICIs）的出现彻底改变了多种癌症的治疗格局。然而，并非所有患者都能从中受益，因此寻找预测ICIs疗效的生物标志物至关重要。AI在此方面发挥着关键作用，通过分析多组学数据（基因组、转录组、表观基因组）、影像组学和数字病理学数据，发现与ICIs疗效相关的预测性生物标志物 73。例如，系统综述指出，80%的AI驱动的生物标志物发现研究发表在2021-2022年，涵盖了基因组学、影像组学、病理组学、真实世界数据和多模态数据，其中非小细胞肺癌和黑色素瘤是研究最多的癌种 73。AI不仅能够识别单一模态的生物标志物，还能整合多模态数据，发现新的“元生物标志物”，为精准免疫肿瘤学提供了新的方向 73。例如，LORIS模型整合临床、病理和基因组特征，能有效预测接受免疫检查点阻断治疗的患者结果，超越了传统标志物的预测能力，为患者分层提供了准确、可解释的方法 74。
- 免疫组库测序分析：单细胞RNA测序结合TCR库分析技术，能够全面揭示肿瘤微环境中T细胞的异质性及其克隆演变。例如，在急性髓系白血病（AML）中，通过对PD-1阻断治疗前后患者骨髓细胞的单细胞RNA分析和TCR库分析，发现治疗响应患者的TCR库扩增主要来自CD8 T细胞，而治疗抵抗患者的TCR库则收缩，揭示了适应性T细胞可塑性和基因组改变在决定PD-1阻断反应中的作用 75。AI的介入使得对这类复杂高维数据的分析变得可行，从而更深入地理解肿瘤免疫应答。
疫苗设计的AI赋能
AI在疫苗设计中扮演着越来越重要的角色，特别是在抗原表位预测、疫苗成分优化和个性化疫苗策略开发方面 76。
- 抗原表位预测与优化：传统疫苗设计通常基于经验，耗时且成本高昂。AI能够通过学习蛋白质序列和结构特征，预测最具免疫原性的T细胞和B细胞表位，从而指导疫苗抗原的Rational设计 77。这对于开发新型传染病疫苗和治疗性肿瘤疫苗尤为重要。
- mRNA疫苗设计：mRNA疫苗在COVID-19大流行中展现出巨大潜力。AI能够优化mRNA序列，提高其稳定性、翻译效率和免疫原性，从而提升mRNA疫苗的效力 77。AI还可以辅助设计编码肿瘤特异性新抗原的个性化mRNA疫苗，这些新抗原由肿瘤细胞特有的基因突变产生，具有高度免疫原性 787980。
- 个性化肿瘤疫苗：AI通过分析患者肿瘤的基因组测序数据，识别出特异性的新抗原，并设计针对这些新抗原的个性化疫苗。这种方法可以激发患者自身的免疫系统，特异性地攻击癌细胞，同时避免对正常组织的损伤 787980。AI在预测新抗原的免疫原性和MHC结合能力方面发挥着核心作用，克服了肿瘤异质性和遗传变异带来的挑战 79。
AI辅助的免疫应答建模与个性化肿瘤免疫治疗
AI不仅在数据分析和疫苗设计中发挥作用，还在建立免疫应答模型和优化个性化肿瘤免疫治疗方案方面展现出巨大潜力。
- 免疫应答建模：系统免疫学（Systems Immunology）结合了实验和计算方法，通过AI对高维免疫数据进行整合分析，构建免疫应答的系统级模型。这有助于揭示保护性免疫的关键机制和通路，例如在流感免疫研究中，AI可以帮助识别长期保护性免疫的关键因素，为更有效的疫苗开发提供支持 81。
- 个性化肿瘤免疫治疗：AI能够综合分析患者的基因组、转录组、蛋白质组以及肿瘤微环境数据，预测患者对不同免疫治疗方案（如免疫检查点抑制剂、CAR-T细胞疗法）的响应 82。通过深度学习模型，可以从医学图像中同时预测肿瘤免疫和基质微环境状态以及治疗结果，从而指导个性化治疗策略 83。例如，AI已被用于预测肺癌患者对免疫疗法的益处和副作用，通过整合放射组学、病理学和基因组学数据，提高PD-L1表达、肿瘤突变负荷（TMB）和肿瘤微环境的预测准确性 82。AI甚至可以识别出对免疫疗法无反应的错配修复缺陷型肿瘤亚群，为联合治疗提供依据 83。
总而言之，AI正在从多个维度推动免疫学研究的进步，不仅提升了对免疫系统复杂性的理解，也加速了新型疫苗和个性化免疫治疗方案的开发，为疾病防控带来了新的希望。
3.2 传染病与抗菌耐药性的AI防控
传染病和抗菌耐药性（AMR）是全球公共卫生面临的严峻挑战。每年，大约有500万人死于与细菌AMR相关的感染84。人工智能（AI）在此领域展现出巨大潜力，通过加速病原体识别、预测耐药模式、建模疫情传播以及支持新药研发，为公共卫生决策和感染控制提供了前所未有的工具和洞察力8586。
AI在病原体识别中的应用
准确快速地识别病原体是有效控制传染病的关键。AI技术正在通过分析基因组数据、形态学特征和临床数据，显著提高病原体识别的效率和准确性8788。
- 基因组数据分析：AI/ML算法能够分析基因组数据，识别与特定病原体相关的遗传标记，甚至区分同一病原体的不同亚种或菌株8589。例如，在快速检测肺炎克雷伯菌（Klebsiella pneumoniae, KP）对头孢他啶-阿维巴坦（CZA）的耐药性方面，结合基质辅助激光解吸电离飞行时间质谱（MALDI-TOF MS）和机器学习（尤其是LightGBM模型）的AI临床决策支持系统（AI-CDSS）展现出高精度（AUC为0.95），比传统方法缩短了检测时间，能更快地指导临床决策90。此外，AI在分析全基因组测序（WGS）数据以进行快速病原体鉴定和抗微生物药物敏感性预测方面也显示出巨大潜力91。
- 图像识别和形态学分析：AI在医学影像分析方面的进展也使其能应用于病原体的可视化检测。例如，AI能够实现显微镜下微生物图像的自动形态识别和图像处理，例如中国科学院微生物研究所开发的一种结合拉曼光谱和AI的单细胞微生物识别技术，实现了高达95.64%的平均准确率，并在5分钟内完成识别92。这对于快速诊断和疫情早期预警具有重要意义。
- 多组学数据整合：新兴的微生物学和感染性疾病研究越来越倾向于整合多组学数据，如基因组学、转录组学等。AI通过整合这些复杂数据，能够识别出高优先级的微生物，为早期发现新发病原体提供支持88。
抗菌耐药性（AMR）预测的AI赋能
抗菌耐药性是全球健康危机，AI/ML被认为是应对AMR的强大工具，能够预测耐药模式、发现新的抗生素，并优化抗微生物药物管理策略848593。
- AMR基因标记挖掘：AI算法能够分析病原体的基因组数据，识别与抗生素耐药性相关的基因标记，从而预测其耐药谱。这使得医生能够更精确地选择抗生素，避免不必要的广谱抗生素使用，从而减缓耐药性的发展8586。例如，通过基因组学研究，可以识别出特定的k-mers（短DNA序列）作为微生物亚种的遗传标记，这对于监测和限制病原体传播的策略至关重要89。
- 耐药性模式预测：机器学习模型可以从大规模临床数据中学习，预测病原体对特定抗生素的敏感性或耐药性。这些模型能够整合患者的临床信息、感染类型和微生物检测结果，为个性化治疗提供依据8587。有评论指出，AI可以辅助抗微生物药物敏感性测试（AST）和AMR监测，从而更好地管理抗生素84。
- 新型抗生素发现：AI正在加速新型抗生素的发现过程。通过虚拟筛选、分子设计和预测化合物的抗菌活性，AI能够显著提高新药研发的效率。例如，AI驱动的从头设计方法可以生成具有抗菌活性的自组装肽材料，这类肽能有效清除生物膜，且不易诱导耐药性94。Genome Mining策略可以用于靶向发现具有抗肿瘤活性的微生物次级代谢产物95。AI已被用于抗微生物药物研发的全过程，包括小分子、抗菌肽、噬菌体疗法和精油的发现9396。
疫情传播建模与公共卫生决策
AI在建模传染病疫情传播、预测爆发趋势和支持公共卫生决策方面发挥着越来越重要的作用97。
- 实时疫情监测与预测：AI模型可以整合多源数据，包括社交媒体信息、移动数据、气候数据和医疗记录，以实时监测疾病传播，并预测潜在的爆发区域和规模。这有助于公共卫生机构及时采取干预措施97。例如，AI模型可以帮助预测新发传染病的传播趋势，为政策制定提供数据支持97。
- 后流行病学研究：在疫情爆发后，AI可以对大量的流行病学数据进行深度分析，揭示疾病传播的关键因素和模式。例如，在流行病学研究中，AI可以分析传播链、识别高风险人群，并评估干预措施的有效性97。AI还能够通过分析基因组、免疫组数据，识别致病过程中的关键免疫病理学过程和控制因子98。
- 抗微生物药物管理（AMS）：AI和ML在抗微生物药物管理方面具有变革性作用，通过数据驱动的干预措施优化抗生素使用，对抗耐药性。AI驱动的预测分析可以识别耐药模式，预测疫情爆发，并通过整合临床和流行病学大数据来指导个性化抗生素治疗86。AI还支持先进诊断工具的开发，减少对广谱抗生素的依赖，并促进及时、有针对性的治疗。
尽管AI在传染病和AMR防控中潜力巨大，但仍面临数据质量、算法可解释性、伦理和隐私等挑战8799。未来的成功将依赖于跨学科合作，开发标准化、公平的AI工具，并加强全球数字基础设施建设，以充分发挥AI在应对这些全球健康挑战中的作用87。
4. 多组学与虚拟生物系统的AI构建
4.1 虚拟细胞与多尺度生物系统建模
传统生物学研究往往难以全面捕捉细胞内部复杂而动态的分子交互网络。随着高通量组学技术和人工智能（AI）的飞速发展，构建“AI虚拟细胞”（AI Virtual Cell, AIVC）已成为可能，它代表了一种多尺度、多模态、基于大型神经网络的模型，旨在模拟和预测分子、细胞和组织在不同状态下的功能和行为 100。这种方法通过整合多组学数据，实现了对细胞功能模拟和疾病机制的深入解析，为药物发现提供了新的视角 101102。
AI虚拟细胞（AIVC）的构建逻辑
AIVC的核心理念是利用AI技术，尤其是深度学习，来处理和整合海量的生物学数据，从而构建一个能够模拟真实细胞行为的计算模型。其构建逻辑主要体现在以下几个方面：
多组学数据整合：AIVC旨在整合细胞内外环境的多种数据流，包括：
多尺度建模：细胞行为是一个多尺度现象，从分子层面的蛋白质相互作用到细胞器层面的功能，再到细胞整体的生理反应。AIVC的目标是实现从分子到细胞乃至组织层面的多尺度建模。这意味着AIVC不仅能模拟单个分子的动态，也能预测整个细胞在不同刺激下的反应 100105。这种多尺度建模能力对于理解复杂疾病的发生发展至关重要，因为疾病往往涉及多个生物学层面的异常 106107。
大型神经网络的应用：AIVC的核心驱动力在于其基于大型神经网络的架构。这些神经网络能够学习数据中复杂的非线性关系，捕捉生物系统固有的复杂性和动态性。深度学习模型能够从非结构化的生物学数据中提取高维特征，并构建预测模型 100108109。
动态模拟与预测能力：与静态模型不同，AIVC具备动态模拟能力，能够预测细胞在时间和空间上的行为变化。例如，通过模拟细胞对药物刺激的响应，AIVC可以预测药物的药效、毒性以及潜在的耐药机制。这种动态模拟对于理解疾病进展、评估治疗效果具有重要意义 100。
细胞功能模拟与疾病机制解析
AIVC的构建将极大地提升我们对细胞功能和疾病机制的理解：
高保真模拟：通过整合多种数据和先进的AI模型，AIVC能够提供高保真的细胞行为模拟。例如，它可以帮助理解癌症的复杂性，通过预测靶向治疗效果来降低临床试验风险 101。这种模拟能力有助于在体外和体内实验之前，对假说进行快速验证和筛选，从而加速研究进程。
复杂疾病机制的揭示：许多疾病，如神经退行性疾病、代谢性疾病和癌症，都涉及复杂的细胞网络失调。AIVC能够系统性地分析这些网络，识别关键的驱动基因、信号通路或分子相互作用，从而深入理解疾病的病理生理学基础。
生物学发现加速：AIVC作为一个计算平台，能够帮助研究人员快速探索生物学问题，生成新的研究假说。它可以通过模拟不同条件下的细胞反应，预测新的生物标志物、潜在的药物靶点或细胞治疗策略，从而加速基础研究和转化研究的发现过程 100。
在药物靶点发现中的应用前景
AIVC在药物靶点发现方面展现出巨大的应用潜力：
虚拟药物筛选与靶点识别：AIVC能够模拟药物分子与细胞内靶点的相互作用，预测其结合亲和力、药效和潜在的副作用。通过在虚拟环境中测试数百万甚至数十亿种化合物，AIVC可以高效地识别出具有治疗潜力的分子，并确定其作用靶点。这与传统的高通量筛选相比，能够大大缩短筛选时间，降低成本。
个性化药物设计：结合患者特异性的组学数据，AIVC可以构建个性化的虚拟细胞模型，模拟不同患者对药物的响应差异。这有助于开发更精准的个性化治疗方案，为患者选择最有效的药物，并预测潜在的不良反应，从而实现精准医疗。
药物再利用（Drug Repurposing）：AIVC能够分析现有药物对细胞网络的影响，从而发现其在治疗其他疾病方面的潜力。通过模拟不同药物在AIVC中的作用，研究人员可以识别出具有新适应症的药物，加速药物的开发和上市。
耐药机制预测与干预策略：许多疾病的治疗面临耐药性问题。AIVC可以通过模拟耐药细胞的分子网络变化，预测耐药发生的机制，并在此基础上设计新的联合用药方案或克服耐药性的策略。
总之，AIVC代表了生物学和医学研究的未来方向。通过整合AI、多组学和多尺度建模，AIVC有望从根本上改变我们对生命过程的理解，加速新药发现和精准医疗的发展，为人类健康带来革命性的影响 100。
4.2 网络药理学与传统医学的AI升级
传统中医药（TCM）以其“多成分、多靶点、多通路”的复杂作用模式治疗疾病，在现代医学中越来越受到关注。然而，其复杂性也给作用机制的阐明带来了挑战。网络药理学（Network Pharmacology, NP）提供了一种从系统层面理解传统药物作用的新范式，而人工智能（AI）技术的融合，正在将传统医学网络药理学（TCM-NP）推向一个全新的精准化和智能化阶段 1314。AI不仅加速了中药有效成分的识别、作用机制的揭示，还优化了配方设计与临床应用，实现了中医药从经验医学向循证医学的转变 15110。
AI驱动的网络关系挖掘
中药复方通常包含数百种化学成分，它们共同作用于多个生物靶点和信号通路，形成复杂的“成分-靶点-疾病”网络。AI技术在挖掘这些复杂的网络关系中发挥着关键作用：
大规模数据集成与知识图谱构建：AI能够整合来自传统中药数据库、化学成分数据库、靶点数据库、基因表达谱、疾病关联数据库等海量异构数据 111。AI可以自动提取中药成分、适应症、药理作用、靶点、通路等信息，并构建结构化的知识图谱。这些知识图谱为后续的网络分析奠定了基础，使得研究人员能够更全面地理解中药的整体作用网络。
复杂网络建模与分析：AI算法，特别是图神经网络（Graph Neural Networks, GNNs）和深度学习模型，能够对构建的复杂生物网络进行建模和分析。例如，通过分析网络拓扑结构，AI可以识别网络中的关键节点（如核心成分、核心靶点）和关键通路。FordNet系统就是一个典型的案例，它融合了表型和分子信息，利用深度学习模型来推荐中药方剂，其性能显著优于传统方法，并能有效学习中医名家的临床经验，标志着中药研究模式从“经验为主、宏观”向“数据驱动、宏观-微观结合”的转变 15。
药效物质基础识别：AI结合色谱-质谱联用技术（如UPLC-QTOF-MS），能够快速鉴定中药中的化学成分，并通过机器学习模型预测这些成分的活性 16。例如，在一项针对“血府逐瘀口服液”的研究中，AI驱动的策略结合多波长融合HPLC指纹图谱和斑马鱼模型评估，发现了七种具有促血管生成活性的化合物，并通过网络药理学和分子对接进一步验证了其功效 112。这大大加速了中药药效物质基础的阐明。
AI赋能的靶点定位与作用机制揭示
AI在中药网络药理学中的另一个核心应用是精准定位药物作用靶点，并深入揭示其分子机制：
高精度靶点预测：AI模型能够利用化合物的化学结构信息和蛋白质的序列/结构信息，预测化合物与靶点之间的结合亲和力。例如，在独活寄生对治疗骨关节炎的研究中，网络药理学结合分子对接策略，识别出槲皮素、白芷醚和β-谷甾醇等核心化合物，并通过PPI网络和GO/KEGG富集分析，揭示了NF-κB、PI3K-AKT和TNF等关键信号通路在治疗中的作用 113。类似地，Qishen Yiqi滴丸治疗糖尿病心力衰竭的研究也通过AI结合网络药理学，确定了108种活性成分和257个潜在靶点，发现其作用机制涉及缺氧反应、活性氧代谢和细胞因子受体相互作用等 114。
多通路协同作用分析：中药治疗复杂疾病往往通过多靶点、多通路的协同作用。AI能够通过网络分析识别这些协同作用模式。例如，葛根芩连汤治疗溃疡性结肠炎的研究通过网络药理学和分子对接，揭示了其抗炎、抗氧化和免疫调节作用，涉及IL-17通路和Th17细胞分化通路等，并确定了槲皮素、山奈酚等关键活性成分 115。AI辅助的临床药理学研究也有助于揭示中药在钾离子通道方面的作用，为抗心律失常药物的开发提供新思路 116。
预测疾病“证候”的分子基础：传统中医的“证候”是疾病在特定阶段的整体表现，具有复杂性和模糊性 117118。AI正在帮助将这些抽象的“证候”转化为可量化的生物学指标。通过对大量临床数据、影像数据、组学数据以及中医“四诊”（望、闻、问、切）信息的学习，AI模型能够建立“证候”与分子机制之间的关联，从而为精准辨证论治提供客观依据。例如，通过深度学习和迁移学习，AI已在舌诊、脉诊和证候分型方面取得显著进展，提高了中医诊断的客观性、标准化和准确性 118。
新型给药策略与质量控制：AI不仅助力机制研究，还在中药的质量控制和现代化生产中发挥作用。例如，结合近红外光谱（NIRS）和AI模型，可以建立快速无损的中药质量控制系统，用于区分过期样品和正常样品，并准确预测活性成分含量，促进中药的标准化和全球化 112。
综上所述，AI与网络药理学的深度融合正在推动传统中医药研究实现从“经验”到“循证”、从“宏观”到“微观”的范式转变，不仅有助于揭示传统药物的复杂作用机制和临床价值，也为新药研发和精准医疗提供了创新工具和方法 1314。
5. 长寿与健康老龄化的AI技术融合
5.1 衰老相关生物标志物的AI发现
衰老是一个复杂的多因素过程，与几乎所有慢性疾病的发生发展密切相关。随着全球预期寿命的延长，如何延长健康寿命（healthspan）而非仅仅生命长度（lifespan）成为研究的焦点。识别准确的衰老生物标志物对于理解衰老机制、评估衰老状态和开发抗衰老干预措施至关重要。人工智能（AI）在处理和整合海量多组学数据方面展现出独特优势，正在革新衰老相关生物标志物的发现范式，尤其是在表观遗传时钟、代谢标志物以及蛋白组学等方面取得了显著进展 119120。
AI驱动的多组学数据挖掘
衰老是一个多尺度的生物学过程，涉及基因组、转录组、表观基因组、蛋白质组和代谢组等多个层面的动态变化。AI技术能够整合这些异构的多组学数据，从中识别出与衰老进程紧密相关的分子特征：
表观遗传时钟（Epigenetic Clocks）：
- DNA甲基化时钟：AI在表观遗传学领域最显著的应用之一是开发基于DNA甲基化的“表观遗传时钟”。这些时钟通过机器学习算法分析特定CpG位点（通常是数万甚至数十万个）的甲基化状态，能够精准预测个体的生物学年龄。例如，Pan Li等人开发的PathwayAge模型，整合了通路层面的DNA甲基化变化，在19个队列的10,615名个体中实现了高精度（Pearson相关系数0.977，平均绝对误差2.350）的年龄预测 121。该模型不仅提高了年龄估计的准确性，还能揭示衰老与疾病之间的机制联系，例如自噬、细胞黏附、突触信号传导和代谢调节等关键通路在衰老中的作用 121。
- 疾病特异性加速衰老：AI模型进一步揭示了表观遗传时钟的加速（Age Acceleration, Age-A）与多种疾病风险的关联。例如，研究发现林奇综合征（Lynch Syndrome）突变携带者可能表现出DNA甲基化时钟的加速 122。基于AI的预测模型，如梯度提升模型，能利用健康检查数据和27个临床因素，高精度地预测生物学年龄（R2达到0.967），并揭示肾功能、血糖、肝功能和人体测量学指标是生物学年龄的重要预测因子 123。尽管环境、生活方式和血管风险因素对Age-A有影响，但即使是最优的机器学习模型（如弹性网络回归和多层感知器）也只能解释其中一小部分变异性，这表明表观遗传时钟的复杂性和多因素驱动性 124。
代谢标志物（Metabolomic Biomarkers）：
蛋白质组标志物（Proteomic Biomarkers）：
- 血浆蛋白质组与健康衰老评分：AI在分析血浆蛋白质组数据中也取得了重要突破。一项针对7,565份血清样本的纵向蛋白质组学研究，在9年随访期内识别出86种与衰老相关的蛋白质，这些蛋白质与32种临床特征和14种主要衰老相关慢性病的发生率有关 125。研究人员利用机器学习模型从这86种蛋白质中筛选出22种，构建了一个“蛋白质组健康衰老评分”（Proteomic Healthy Ageing Score, PHAS），该评分能够预测心血管代谢疾病的发生率 125。此外，AI模型还发现肠道微生物群是影响PHAS的可调节因素，为通过调节微生物群促进健康衰老提供了潜在干预靶点 125。
- 脑脊液蛋白质组时钟：AI也被用于分析脑脊液（CSF）蛋白质组数据，以创建预测年龄的“生物学时钟”，从而深入了解大脑衰老机制。一项研究开发了一个基于CSF蛋白质组的机器学习时钟，能够以0.79的Pearson相关系数和4.30年的平均绝对误差预测个体的实际年龄，并能通过少量蛋白质特征实现相似的准确性 126。这个时钟揭示了在与其他蛋白质相互作用时，对年龄预测具有高度影响但自身与年龄不直接相关的蛋白质，为中枢神经系统衰老机制提供了新的洞见 126。
健康寿命预测模型与衰老机制解析
AI不仅能预测生物学年龄，还能基于多组学数据构建健康寿命预测模型，并揭示衰老的非线性动态和器官特异性衰老模式：
非线性衰老动态：传统研究常侧重于衰老的线性变化，但AI驱动的多组学分析揭示了人类衰老的非线性特征。一项针对108名25至75岁参与者的长期多组学研究发现，衰老相关分子标志物的显著失调主要发生在约44岁和60岁这两个关键时期 127。不同分子和功能通路与这些时期相关，例如免疫调节和碳水化合物代谢在60岁左右发生变化，而心血管疾病、脂质和酒精代谢变化在40岁左右出现 127。这项研究强调了衰老相关疾病的功能和风险在人类寿命中呈非线性变化，为理解衰老的分子和生物学通路提供了新见解 127。
器官特异性衰老：AI促进了“年龄型（ageotype）”概念的发展，即个体独特的衰老轨迹以及不同器官的异步衰老过程 128129。例如，对人类卵巢衰老的研究结合单细胞RNA测序和空间转录组学，系统地描绘了卵巢细胞在衰老过程中的时空转录组变化，发现DNA损伤反应可能是卵母细胞衰老的关键通路 130。FOXP1被鉴定为卵巢衰老的调节因子，其表达随年龄下降并抑制CDKN1A转录，沉默FOXP1会导致小鼠卵巢早衰 130。这些发现为诊断生物标志物和治疗策略的优先级排序提供了全面理解 130。AI能够整合临床、生理、行为和多组学参数，构建生物学年龄评分来反映不同的生理系统，揭示了性别特异性的衰老模式，并发现较高的生物学年龄评分与年龄相关疾病的较高患病率相关联 128。这些研究成果为实现个性化的抗衰老医学奠定了基础 128129。
综上所述，AI在衰老生物标志物的发现和健康寿命预测中发挥着越来越关键的作用。通过整合和分析多组学数据，AI不仅能够提供更精准的生物学年龄评估，还能深入揭示衰老过程中的分子机制、非线性动态和器官特异性变化，为理解衰老本质和开发有效的抗衰老策略提供了前所未有的工具和洞察力。
5.2 抗衰老干预的AI辅助开发
随着对衰老机制理解的深入，开发有效的抗衰老干预措施已成为生物科技领域的重要方向。人工智能（AI）正在这一进程中扮演越来越关键的角色，尤其是在抗衰老药物的筛选、临床干预方案的设计以及加速基础研究到临床转化的全链条。AI通过其强大的数据分析、模式识别和预测能力，显著提升了抗衰老研究的效率和成功率 131132。
AI在抗衰老药物筛选中的应用
传统的药物筛选过程耗时且成本高昂，而AI的引入极大地加速了潜在抗衰老化合物的发现。
高通量虚拟筛选与分子设计：AI模型能够对数百万甚至数十亿种化合物进行虚拟筛选，预测其与已知衰老相关靶点（如mTOR通路、NAD+代谢途径、细胞衰老相关通路）的结合亲和力及药理活性。例如，通过机器学习算法训练现有数据，AI能够以数百倍的成本降低效率发现新的衰老溶解剂（senolytics）133。一项研究利用机器学习模型，通过分析已发表数据，成功发现了三种具有与已知衰老溶解剂相当效力的化合物：银杏素（ginkgetin）、夹竹桃苷（periplocin）和夹竹桃苦苷（oleandrin）133。其中，银杏素被进一步证实能通过靶向STING通路来缓解炎症和细胞衰老，这一发现结合了AI算法与药理学方法，为银杏叶提取物的抗衰老作用提供了理论基础 134。
组合药物与协同作用预测：衰老是一个多因素过程，单一药物往往难以完全逆转其影响。AI在预测药物组合的协同抗衰老效应方面具有优势。通过分析不同药物作用于衰老相关基因表达模式的数据，AI系统（如基于深度神经网络，DNN）能够识别出具有更强抗衰老效果的药物组合 135。这种方法通过识别药物对差分表达基因（DEG）的影响，相比单一疗法能更有效地延缓衰老效应 135。
新型抗衰老肽的发现：AI结合计算生物学，可以从复杂的天然产物或再生细胞分泌组中快速筛选具有抗衰老潜力的活性分子。例如，通过虚拟筛选，研究人员从再生细胞分泌组中鉴定出一种名为EQ-9的新型功能性肽，该肽在体外实验中展现出显著的抗衰老和皮肤修复效果 136。这一范式有望应用于更广泛的天然资源，以发现针对特定生物学功能的新型肽分子 136。
AI辅助的临床干预方案设计
AI不仅限于药物发现，还在优化临床试验设计、个性化抗衰老策略以及管理衰老相关疾病方面发挥作用。
临床试验优化与患者分层：AI可以分析临床数据、生物标志物和基因组信息，帮助识别对特定抗衰老干预措施响应最佳的患者亚群 131。这有助于设计更高效、更有针对性的临床试验，加速抗衰老疗法的开发。例如，AI在靶点识别中发挥日益重要的作用，它能够分析大型数据集和复杂的生物网络，帮助研究人员平衡靶点选择的新颖性和置信度 137。
精准衰老评估与健康寿命预测：AI模型能够利用多组学数据（如基因组学、转录组学、代谢组学、表观遗传组学）构建更精准的生物学年龄时钟和健康寿命预测模型 131132138。例如，ENABL Age框架结合机器学习和可解释AI（XAI）方法，通过血检或问卷数据准确估计生物学年龄，并提供个体化的风险因素解释，其预测死亡率的能力优于现有模型 139。这些工具可以帮助医生识别加速衰老的个体，并为其制定个性化的干预策略 139140。
抗衰老干预的个性化：考虑到个体遗传背景、生活方式和环境因素的差异，个性化抗衰老干预至关重要。AI能够整合这些信息，推荐最适合个体的营养、运动和药物干预方案。例如，通过整合多种生物标志物和“组学”技术，AI可以帮助实现更个性化的抗衰老策略，从分子层面理解衰老，从而开发针对性的干预措施 138141。
长寿生物技术案例与转化加速
AI正在推动长寿生物技术从基础研究向临床应用的快速转化 131132。
靶向衰老机制的药物开发：AI已被用于识别和优化针对衰老标志（Hallmarks of Aging）的药物，例如细胞衰老、炎症（inflammaging）、蛋白质稳态失衡等 132141。通过机器学习，研究人员能够从海量数据中挖掘出新的抗衰老靶点和作用机制，并开发出更具特异性和有效性的药物。AI甚至可以帮助发现新的抗衰老疗法，如靶向丙酸代谢途径的药物，因为该途径在肌肉衰老中表现出差异分布 142。
与癌症治疗的交叉应用：衰老和癌症之间存在复杂的关联。AI在癌症药物开发中的应用，特别是在理解衰老对癌症治疗反应的影响方面，也为抗衰老研究提供了借鉴 143。通过分析衰老对免疫系统、肿瘤微环境和药物代谢的影响，AI可以帮助优化针对老年癌症患者的治疗方案，间接促进健康老龄化。
开放科学与数据共享的推动：AI驱动的抗衰老研究鼓励数据共享，通过汇集全球研究数据和计算资源，共同加速抗衰老疗法的发现和验证。这种模式有助于克服传统研究的局限性，更快地将实验室发现转化为临床实践。
综上所述，AI正在多方面赋能抗衰老研究，从高效药物筛选到精准临床干预，再到加速技术转化，为实现健康长寿的目标提供了前所未有的机遇 131132。随着AI技术的不断发展和生物学数据的日益丰富，我们有理由相信，在不久的将来，AI将在延长人类健康寿命方面发挥更加决定性的作用 132144。
内容由 AI 生成，仅供参考，请仔细甄别
参考文献
1Artificial intelligence to deep learning: machine intelligence approach for drug discovery.PubMed
Rohan Gupta, Devesh Srivastava, Mehar Sahu, et al.
Mol Divers. 2021 Aug;25(3):1315-1360. doi: 10.1007/s11030-021-10217-3. Epub 2021 Apr 12.
Drug designing and development is an important area of research for pharmaceutical companies and chemical scientists. However, low efficacy, off-target delivery, time consumption, and high cost impose a hurdle and challenges that impact drug design and discovery. Further, complex and big data from genomics, proteomics, microarray data, and clinical trials also impose an obstacle in the drug discovery pipeline. Artificial intelligence and machine learning technology play a crucial role in drug discovery and development. In other words, artificial neural networks and deep learning algorithms have modernized the area. Machine learning and deep learning algorithms have been implemented in several drug discovery processes such as peptide synthesis, structure-based virtual screening, ligand-based virtual screening, toxicity prediction, drug monitoring and release, pharmacophore modeling, quantitative structure-activity relationship, drug repositioning, polypharmacology, and physiochemical activity. Evidence from the past strengthens the implementation of artificial intelligence and deep learning in this field. Moreover, novel data mining, curation, and management techniques provided critical support to recently developed modeling algorithms. In summary, artificial intelligence and deep learning advancements provide an excellent opportunity for rational drug design and discovery process, which will eventually impact mankind. The primary concern associated with drug design and development is time consumption and production cost. Further, inefficiency, inaccurate target delivery, and inappropriate dosage are other hurdles that inhibit the process of drug delivery and development. With advancements in technology, computer-aided drug design integrating artificial intelligence algorithms can eliminate the challenges and hurdles of traditional drug design and development. Artificial intelligence is referred to as superset comprising machine learning, whereas machine learning comprises supervised learning, unsupervised learning, and reinforcement learning. Further, deep learning, a subset of machine learning, has been extensively implemented in drug design and development. The artificial neural network, deep neural network, support vector machines, classification and regression, generative adversarial networks, symbolic learning, and meta-learning are examples of the algorithms applied to the drug design and discovery process. Artificial intelligence has been applied to different areas of drug design and development process, such as from peptide synthesis to molecule design, virtual screening to molecular docking, quantitative structure-activity relationship to drug repositioning, protein misfolding to protein-protein interactions, and molecular pathway identification to polypharmacology. Artificial intelligence principles have been applied to the classification of active and inactive, monitoring drug release, pre-clinical and clinical development, primary and secondary drug screening, biomarker development, pharmaceutical manufacturing, bioactivity identification and physiochemical properties, prediction of toxicity, and identification of mode of action.
2Artificial Intelligence and Machine Learning Technology Driven Modern Drug Discovery and Development.PubMed
Chayna Sarkar, Biswadeep Das, Vikram Singh Rawat, et al.
Int J Mol Sci. 2023 Jan 19;24(3):2026. doi: 10.3390/ijms24032026.
The discovery and advances of medicines may be considered as the ultimate relevant translational science effort that adds to human invulnerability and happiness. But advancing a fresh medication is a quite convoluted, costly, and protracted operation, normally costing USD ~2.6 billion and consuming a mean time span of 12 years. Methods to cut back expenditure and hasten new drug discovery have prompted an arduous and compelling brainstorming exercise in the pharmaceutical industry. The engagement of Artificial Intelligence (AI), including the deep-learning (DL) component in particular, has been facilitated by the employment of classified big data, in concert with strikingly reinforced computing prowess and cloud storage, across all fields. AI has energized computer-facilitated drug discovery. An unrestricted espousing of machine learning (ML), especially DL, in many scientific specialties, and the technological refinements in computing hardware and software, in concert with various aspects of the problem, sustain this progress. ML algorithms have been extensively engaged for computer-facilitated drug discovery. DL methods, such as artificial neural networks (ANNs) comprising multiple buried processing layers, have of late seen a resurgence due to their capability to power automatic attribute elicitations from the input data, coupled with their ability to obtain nonlinear input-output pertinencies. Such features of DL methods augment classical ML techniques which bank on human-contrived molecular descriptors. A major part of the early reluctance concerning utility of AI in pharmaceutical discovery has begun to melt, thereby advancing medicinal chemistry. AI, along with modern experimental technical knowledge, is anticipated to invigorate the quest for new and improved pharmaceuticals in an expeditious, economical, and increasingly compelling manner. DL-facilitated methods have just initiated kickstarting for some integral issues in drug discovery. Many technological advances, such as "message-passing paradigms", "spatial-symmetry-preserving networks", "hybrid de novo design", and other ingenious ML exemplars, will definitely come to be pervasively widespread and help dissect many of the biggest, and most intriguing inquiries. Open data allocation and model augmentation will exert a decisive hold during the progress of drug discovery employing AI. This review will address the impending utilizations of AI to refine and bolster the drug discovery operation.
3Unleashing the power of generative AI in drug discovery.PubMed
Amit Gangwal, Antonio Lavecchia
Drug Discov Today. 2024 Jun;29(6):103992. doi: 10.1016/j.drudis.2024.103992. Epub 2024 Apr 23.
Artificial intelligence (AI) is revolutionizing drug discovery by enhancing precision, reducing timelines and costs, and enabling AI-driven computer-aided drug design. This review focuses on recent advancements in deep generative models (DGMs) for de novo drug design, exploring diverse algorithms and their profound impact. It critically analyses the challenges that are intricately interwoven into these technologies, proposing strategies to unlock their full potential. It features case studies of both successes and failures in advancing drugs to clinical trials with AI assistance. Last, it outlines a forward-looking plan for optimizing DGMs in de novo drug design, thereby fostering faster and more cost-effective drug development.
4Artificial intelligence in drug design.PubMed
Feisheng Zhong, Jing Xing, Xutong Li, et al.
Sci China Life Sci. 2018 Oct;61(10):1191-1204. doi: 10.1007/s11427-018-9342-2. Epub 2018 Jul 18.
Thanks to the fast improvement of the computing power and the rapid development of the computational chemistry and biology, the computer-aided drug design techniques have been successfully applied in almost every stage of the drug discovery and development pipeline to speed up the process of research and reduce the cost and risk related to preclinical and clinical trials. Owing to the development of machine learning theory and the accumulation of pharmacological data, the artificial intelligence (AI) technology, as a powerful data mining tool, has cut a figure in various fields of the drug design, such as virtual screening, activity scoring, quantitative structure-activity relationship (QSAR) analysis, de novo drug design, and in silico evaluation of absorption, distribution, metabolism, excretion and toxicity (ADME/T) properties. Although it is still challenging to provide a physical explanation of the AI-based models, it indeed has been acting as a great power to help manipulating the drug discovery through the versatile frameworks. Recently, due to the strong generalization ability and powerful feature extraction capability, deep learning methods have been employed in predicting the molecular properties as well as generating the desired molecules, which will further promote the application of AI technologies in the field of drug design.
5Artificial Intelligence, Big Data and Machine Learning Approaches in Precision Medicine & Drug Discovery.PubMed
Anuraj Nayarisseri, Ravina Khandelwal, Poonam Tanwar, et al.
Curr Drug Targets. 2021;22(6):631-655. doi: 10.2174/1389450122999210104205732.
Artificial Intelligence revolutionizes the drug development process that can quickly identify potential biologically active compounds from millions of candidate within a short period. The present review is an overview based on some applications of Machine Learning based tools, such as GOLD, Deep PVP, LIB SVM, etc. and the algorithms involved such as support vector machine (SVM), random forest (RF), decision tree and Artificial Neural Network (ANN), etc. at various stages of drug designing and development. These techniques can be employed in SNP discoveries, drug repurposing, ligand-based drug design (LBDD), Ligand-based Virtual Screening (LBVS) and Structure- based Virtual Screening (SBVS), Lead identification, quantitative structure-activity relationship (QSAR) modeling, and ADMET analysis. It is demonstrated that SVM exhibited better performance in indicating that the classification model will have great applications on human intestinal absorption (HIA) predictions. Successful cases have been reported which demonstrate the efficiency of SVM and RF models in identifying JFD00950 as a novel compound targeting against a colon cancer cell line, DLD-1, by inhibition of FEN1 cytotoxic and cleavage activity. Furthermore, a QSAR model was also used to predict flavonoid inhibitory effects on AR activity as a potent treatment for diabetes mellitus (DM), using ANN. Hence, in the era of big data, ML approaches have been evolved as a powerful and efficient way to deal with the huge amounts of generated data from modern drug discovery to model small-molecule drugs, gene biomarkers and identifying the novel drug targets for various diseases.
6Machine Learning-Enabled Drug-Induced Toxicity Prediction.PubMed
Changsen Bai, Lianlian Wu, Ruijiang Li, et al.
Adv Sci (Weinh). 2025 Apr;12(16):e2413405. doi: 10.1002/advs.202413405. Epub 2025 Feb 3.
Unexpected toxicity has become a significant obstacle to drug candidate development, accounting for 30% of drug discovery failures. Traditional toxicity assessment through animal testing is costly and time-consuming. Big data and artificial intelligence (AI), especially machine learning (ML), are robustly contributing to innovation and progress in toxicology research. However, the optimal AI model for different types of toxicity usually varies, making it essential to conduct comparative analyses of AI methods across toxicity domains. The diverse data sources also pose challenges for researchers focusing on specific toxicity studies. In this review, 10 categories of drug-induced toxicity is examined, summarizing the characteristics and applicable ML models, including both predictive and interpretable algorithms, striking a balance between breadth and depth. Key databases and tools used in toxicity prediction are also highlighted, including toxicology, chemical, multi-omics, and benchmark databases, organized by their focus and function to clarify their roles in drug-induced toxicity prediction. Finally, strategies to turn challenges into opportunities are analyzed and discussed. This review may provide researchers with a valuable reference for understanding and utilizing the available resources to bridge prediction and mechanistic insights, and further advance the application of ML in drugs-induced toxicity prediction.
7Artificial Intelligence in Drug Toxicity Prediction: Recent Advances, Challenges, and Future Perspectives.PubMed
Thi Tuyet Van Tran, Agung Surya Wibowo, Hilal Tayara, et al.
J Chem Inf Model. 2023 May 8;63(9):2628-2643. doi: 10.1021/acs.jcim.3c00200. Epub 2023 Apr 26.
Toxicity prediction is a critical step in the drug discovery process that helps identify and prioritize compounds with the greatest potential for safe and effective use in humans, while also reducing the risk of costly late-stage failures. It is estimated that over 30% of drug candidates are discarded owing to toxicity. Recently, artificial intelligence (AI) has been used to improve drug toxicity prediction as it provides more accurate and efficient methods for identifying the potentially toxic effects of new compounds before they are tested in human clinical trials, thus saving time and money. In this review, we present an overview of recent advances in AI-based drug toxicity prediction, including the use of various machine learning algorithms and deep learning architectures, of six major toxicity properties and Tox21 assay end points. Additionally, we provide a list of public data sources and useful toxicity prediction tools for the research community and highlight the challenges that must be addressed to enhance model performance. Finally, we discuss future perspectives for AI-based drug toxicity prediction. This review can aid researchers in understanding toxicity prediction and pave the way for new methods of drug discovery.
8MolToxPred: small molecule toxicity prediction using machine learning approach.PubMed
Anjali Setiya, Vinod Jani, Uddhavesh Sonavane, et al.
RSC Adv. 2024 Jan 30;14(6):4201-4220. doi: 10.1039/d3ra07322j. eCollection 2024 Jan 23.
Different types of chemicals and products may exhibit various health risks when administered into the human body. For toxicity reasons, the number of new drugs entering the market through the conventional drug development process has been reduced over the years. However, with the advent of big data and artificial intelligence, machine learning techniques have emerged as a potential solution for predicting toxicity and ensuring efficient drug development and chemical safety. An ML model for toxicity prediction can reduce experimental costs and time while addressing ethical concerns by drastically reducing the need for animals and clinical trials. Herein, MolToxPred, an ML-based tool, has been developed using a stacked model approach to predict the potential toxicity of small molecules and metabolites. The stacked model consists of random forest, multi-layer perceptron, and LightGBM as base classifiers and Logistic Regression as the meta classifier. For training and validation purposes, a comprehensive set of toxic and non-toxic molecules is curated. Different structural and physicochemical-based features in the form of molecular descriptors and fingerprints were employed. MolToxPred utilizes a comprehensive feature selection process and optimizes its hyperparameters through Bayesian optimization with stratified 5-fold cross-validation. In the evaluation phase, MolToxPred achieved an AUROC of 87.76% on the test set and 88.84% on an external validation set. The McNemar test was used as the post-hoc test to determine if the stacked models' performance was significantly different compared to the base learners. The developed stacked model outperformed its base classifiers and an existing tool in the literature, reaffirming its better performance. The hypothesis is that the incorporation of a diverse set of data, the subsequent feature selection, and a stacked ensemble approach give MolToxPred the edge over other methods. In addition to this, an attempt has been made to identify structural alerts responsible for endpoints of the Tox21 data to determine the association of a molecule with a plausible downstream pathway of action. MolToxPred may be helpful for drug discovery and regulatory pipelines in pharmaceutical and other industries for toxicity prediction of small molecule candidates.
9Image Based Liver Toxicity Prediction.PubMed
Ece Asilar, Jennifer Hemmerich, Gerhard F Ecker
J Chem Inf Model. 2020 Mar 23;60(3):1111-1121. doi: 10.1021/acs.jcim.9b00713. Epub 2020 Feb 7.
The drugs we use to cure our diseases can cause damage to the liver as it is the primary organ responsible for metabolism of environmental chemicals and drugs. To identify and eliminate potentially problematic drug candidates in the early stages of drug discovery, in silico techniques provide quick and practical solutions for toxicity determination. Deep learning has emerged as one of the solutions in recent years in the field of pharmaceutical chemistry. Generally, in the case of small data sets as used in toxicology, these data-hungry algorithms are prone to overfitting. We approach the problem from two sides. First, we use images of the three-dimensional conformations and benefit from convolutional neural networks which have fewer parameters than the standard deep neural networks with similar depth. Using images allows connecting various chemical features to the geometry of the compounds. Second, we employ the method COVER to up-sample the data set. It is used not only for increasing the size of the data set, but also for balancing the two classes, i.e., toxic and not toxic. The proof of concept is performed on the p53 end point from the Tox21 data set. The results, which are compatible with the winners of the data challenge, encouraged us to use our methods to predict liver toxicity. We use the most extensive publicly available liver toxicity data set by Mulliner et al. and obtain a sensitivity of 0.79 and a specificity of 0.52. These results demonstrate the applicability of image based toxicity prediction using deep neural networks.
10Artificial intelligence for natural product drug discovery.PubMed
Michael W Mullowney, Katherine R Duncan, Somayah S Elsayed, et al.
Nat Rev Drug Discov. 2023 Nov;22(11):895-916. doi: 10.1038/s41573-023-00774-7. Epub 2023 Sep 11.
Developments in computational omics technologies have provided new means to access the hidden diversity of natural products, unearthing new potential for drug discovery. In parallel, artificial intelligence approaches such as machine learning have led to exciting developments in the computational drug design field, facilitating biological activity prediction and de novo drug design for molecular targets of interest. Here, we describe current and future synergies between these developments to effectively identify drug candidates from the plethora of molecules produced by nature. We also discuss how to address key challenges in realizing the potential of these synergies, such as the need for high-quality datasets to train deep learning algorithms and appropriate strategies for algorithm validation.
11The strategies and techniques of drug discovery from natural products.PubMed
Li Zhang, Junke Song, Linglei Kong, et al.
Pharmacol Ther. 2020 Dec;216:107686. doi: 10.1016/j.pharmthera.2020.107686. Epub 2020 Sep 19.
Natural products have been the main sources of new drugs. The different strategies have been developed to find the new drugs based on natural products. The traditional and ethic medicines have provided information on the therapeutic effects and resulted in some notable drug discovery of natural products. The special activities of the medicine plants such as the side effects have inspired scientists to develop the novel small molecular. The microorganisms and the endogenous active substances from human or animal also become the important approaches to the drug discovery. The tremendous progress in technology led to the new strategies in drug discovery from natural products. The bioinformation and artificial intelligence have facilitated the research and development of natural products. We will provide a scene of strategies and technologies for drug discovery from natural products in this review.
12DerivaPredict: A User-Friendly Tool for Predicting and Evaluating Active Derivatives of Natural Products.PubMed
Yu Song, Meng Zhang, Sihao Chang, et al.
Molecules. 2025 Apr 9;30(8):1683. doi: 10.3390/molecules30081683.
While natural products and derivatives have been crucial in drug discovery, the current databases are limited to known compounds. There is a need for tools that can automatically generate and assess novel derivatives of natural products to enhance early-stage drug discovery. We present DerivaPredict (v1.0), a user-friendly tool that generates novel natural product derivatives through chemical and metabolic transformations. It predicts binding affinities using pretrained deep learning models and assesses drug-likeness via ADMET profiling. DerivaPredict is freely accessible with a source code on GitHub.
13Network pharmacology: towards the artificial intelligence-based precision traditional Chinese medicine.PubMed
Peng Zhang, Dingfan Zhang, Wuai Zhou, et al.
Brief Bioinform. 2023 Nov 22;25(1). doi: 10.1093/bib/bbad518.
Network pharmacology (NP) provides a new methodological perspective for understanding traditional medicine from a holistic perspective, giving rise to frontiers such as traditional Chinese medicine network pharmacology (TCM-NP). With the development of artificial intelligence (AI) technology, it is key for NP to develop network-based AI methods to reveal the treatment mechanism of complex diseases from massive omics data. In this review, focusing on the TCM-NP, we summarize involved AI methods into three categories: network relationship mining, network target positioning and network target navigating, and present the typical application of TCM-NP in uncovering biological basis and clinical value of Cold/Hot syndromes. Collectively, our review provides researchers with an innovative overview of the methodological progress of NP and its application in TCM from the AI perspective.
14Network pharmacology approaches for research of Traditional Chinese Medicines.PubMed
Xiang Li, Ziqi Liu, Jie Liao, et al.
Chin J Nat Med. 2023 May;21(5):323-332. doi: 10.1016/S1875-5364(23)60429-7.
Pharmacodynamics material basis and effective mechanisms are the two main issues to decipher the mechnisms of action of Traditional Chinese medicines (TCMs) for the treatment of diseases. TCMs, in "multi-component, multi-target, multi-pathway" paradigm, show satisfactory clinical results in complex diseases. New ideas and methods are urgently needed to explain the complex interactions between TCMs and diseases. Network pharmacology (NP) provides a novel paradigm to uncover and visualize the underlying interaction networks of TCMs against multifactorial diseases. The development and application of NP has promoted the safety, efficacy, and mechanism investigations of TCMs, which then reinforces the credibility and popularity of TCMs. The current organ-centricity of medicine and the "one disease-one target-one drug" dogma obstruct the understanding of complex diseases and the development of effective drugs. Therefore, more attentions should be paid to shift from "phenotype and symptom" to "endotype and cause" in understanding and redefining current diseases. In the past two decades, with the advent of advanced and intelligent technologies (such as metabolomics, proteomics, transcriptomics, single-cell omics, and artificial intelligence), NP has been improved and deeply implemented, and presented its great value and potential as the next drug-discovery paradigm. NP is developed to cure causal mechanisms instead of treating symptoms. This review briefly summarizes the recent research progress on NP application in TCMs for efficacy research, mechanism elucidation, target prediction, safety evaluation, drug repurposing, and drug design.
15FordNet: Recommending traditional Chinese medicine formula via deep neural network integrating phenotype and molecule.PubMed
Wuai Zhou, Kuo Yang, Jianyang Zeng, et al.
Pharmacol Res. 2021 Nov;173:105752. doi: 10.1016/j.phrs.2021.105752. Epub 2021 Sep 2.
Traditional Chinese medicine (TCM) formula is widely used for thousands of years in clinical practice. With the development of artificial intelligence, deep learning models may help doctors prescribe reasonable formulas. Meanwhile, current studies of formula recommendation only focus on the observable clinical symptoms and lack of molecular information. Here, inspired by the theory of TCM network pharmacology, we propose an intelligent formula recommendation system based on deep learning (FordNet), fusing the information of phenotype and molecule. We collected more than 20,000 electronic health records from TCM Master Li Jiren's experience from 2013 to March 2020. In the FordNet system, the feature of diagnosis description is extracted by convolution neural network and the feature of TCM formula is extracted by network embedding, which fusing the molecular information. A hierarchical sampling strategy for data augmentation is designed to effectively learn training samples. Based on the expanded samples, a deep neural network based quantitative optimization model is developed for TCM formula recommendation. FordNet performs significantly better than baseline methods (hit ratio of top 10 improved by 46.9% compared with the best baseline random forest method). Moreover, the molecular information helps FordNet improve 17.3% hit ratio compared with the model using only macro information. Clinical evaluation shows that FordNet can well learn the effective experience of TCM Master and obtain excellent recommendation results. Our study, for the first time, proposes an intelligent recommendation system for TCM formula integrating phenotype and molecule information, which has potential to improve clinical diagnosis and treatment, and promote the shift of TCM research pattern from "experience based, macro" to "data based, macro-micro combined" as well as the development of TCM network pharmacology.
16Machine Learning-Enhanced Network Pharmacology in Traditional Chinese Medicine: Mechanistic Insights Into Chai Hu Gui Zhi Tang for Allergic Rhinitis.PubMed
Qi Yang, Jiageng Guo, Haoqiong Lin, et al.
Chem Biodivers. 2025 Sep;22(9):e202500214. doi: 10.1002/cbdv.202500214. Epub 2025 Apr 22.
Network pharmacology has become a widely used approach for studying complex herbal medicines. This method helps researchers identify potentially active compounds and mechanisms, which can then guide further experiments. However, current network pharmacology methods often face issues like inconsistent compound records and unreliable screening standards, leading to inaccurate results. To address these challenges, we developed an improved workflow using Chai Hu Gui Zhi Tang (CHGZT) as an example. First, we used advanced analytical techniques (ultra-high-performance liquid chromatography quadrupole time-of-flight mass spectrometry) to rapidly identify chemical components in the herbal formula. Next, we created a machine learning model to predict compounds with anti-allergic rhinitis activity, allowing systematic selection of key components for network analysis. Our results showed that specific compounds like cinnamic acid and citric acid may combat allergic rhinitis by regulating immune-related genes (interleukin [IL]-4 and IL-5) while influencing biological processes such as "stress response" and "metabolism of foreign substances." These findings confirm the effectiveness of our optimized method and highlight CHGZT's potential as a therapeutic option for allergic rhinitis.
17Artificial Intelligence Applied to clinical trials: opportunities and challenges.PubMed
Scott Askin, Denis Burkhalter, Gilda Calado, et al.
Health Technol (Berl). 2023;13(2):203-213. doi: 10.1007/s12553-023-00738-2. Epub 2023 Feb 28.
BACKGROUND: Clinical Trials (CTs) remain the foundation of safe and effective drug development. Given the evolving data-driven and personalized medicine approach in healthcare, it is imperative for companies and regulators to utilize tailored Artificial Intelligence (AI) solutions that enable expeditious and streamlined clinical research. In this paper, we identified opportunities, challenges, and potential implications of AI in CTs. METHODS: Following an extensive search in relevant databases and websites, we gathered publications tackling the use of AI and Machine Learning (ML) in CTs from the past 5 years in the US and Europe, including Regulatory Authorities' documents. RESULTS: Documented applications of AI commonly concern the oncology field and are mostly being applied in the area of recruitment. Main opportunities discussed aim to create efficiencies across CT activities, including the ability to reduce sample sizes, improve enrollment and conduct faster, more optimized adaptive CTs. While AI is an area of enthusiastic development, the identified challenges are ethical in nature and relate to data availability, standards, and most importantly, lack of regulatory guidance hindering the acceptance of AI tools in drug development. However, future implications are significant and are anticipated to improve the probability of success, reduce trial burden and overall, speed up research and regulatory approval. CONCLUSION: The use of AI in CTs is in its relative infancy; however, it is a fast-evolving field. As regulators provide more guidance on the acceptability of AI in specific areas, we anticipate the scope of use to broaden and the volume of implementation to increase rapidly.
18Recent innovations in clinical trial design for inflammatory bowel disease.PubMed
Rocio Sedano, Christopher Ma, Vipul Jairath
Curr Opin Pharmacol. 2025 Oct;84:102551. doi: 10.1016/j.coph.2025.102551. Epub 2025 Jul 5.
Clinical trial design in inflammatory bowel disease (IBD) is evolving to address challenges in drug development and approvals. For clinical development, notable innovations include Bayesian designs, adaptive designs, integrated-phase trials and master protocols (such as umbrella, basket, and platform trials). The inclusion of biomarker-driven strategies and precision medicine (PM) trials bring aim to enable patient stratification based on prognostic or predictive markers, leveraging molecular signatures to customize therapy. However, recent studies highlight both the promise and complexity of this approach. Patient-reported outcomes (PROs) have gained prominence as key endpoints, aligning trials with patient-centric measures and regulatory guidance that emphasize symptoms and quality-of-life metrics. Digital health tools and artificial intelligence (AI) are being integrated to streamline trial conduct, from remote monitoring and telemedicine visits to AI-assisted recruitment and data analysis. Pragmatic trials and the integration of real-world evidence (RWE) aim to complement traditional efficacy trials by evaluating treatments in routine care settings. Together, these innovations mark a new era in IBD clinical trial design, aiming to expedite therapeutic development and enhance the relevance of trials to patient care.
19Clinical trial design in the era of precision medicine.PubMed
Elena Fountzilas, Apostolia M Tsimberidou, Henry Hiep Vo, et al.
Genome Med. 2022 Aug 31;14(1):101. doi: 10.1186/s13073-022-01102-1.
Recent rapid biotechnological breakthroughs have led to the identification of complex and unique molecular features that drive malignancies. Precision medicine has exploited next-generation sequencing and matched targeted therapy/immunotherapy deployment to successfully transform the outlook for several fatal cancers. Tumor and liquid biopsy genomic profiling and transcriptomic, immunomic, and proteomic interrogation can now all be leveraged to optimize therapy. Multiple new trial designs, including basket and umbrella trials, master platform trials, and N-of-1 patient-centric studies, are beginning to supplant standard phase I, II, and III protocols, allowing for accelerated drug evaluation and approval and molecular-based individualized treatment. Furthermore, real-world data, as well as exploitation of digital apps and structured observational registries, and the utilization of machine learning and/or artificial intelligence, may further accelerate knowledge acquisition. Overall, clinical trials have evolved, shifting from tumor type-centered to gene-directed and histology-agnostic trials, with innovative adaptive designs and personalized combination treatment strategies tailored to individual biomarker profiles. Some, but not all, novel trials now demonstrate that matched therapy correlates with superior outcomes compared to non-matched therapy across tumor types and in specific cancers. To further improve the precision medicine paradigm, the strategy of matching drugs to patients based on molecular features should be implemented earlier in the disease course, and cancers should have comprehensive multi-omic (genomics, transcriptomics, proteomics, immunomic) tumor profiling. To overcome cancer complexity, moving from drug-centric to patient-centric individualized combination therapy is critical. This review focuses on the design, advantages, limitations, and challenges of a spectrum of clinical trial designs in the era of precision oncology.
21Explainable AI enables clinical trial patient selection to retrospectively improve treatment effects in schizophrenia.PubMed
Monika S Mellem, Matt Kollada, Jane Tiller, et al.
BMC Med Inform Decis Mak. 2021 May 20;21(1):162. doi: 10.1186/s12911-021-01510-0.
BACKGROUND: Heterogeneity among patients' responses to treatment is prevalent in psychiatric disorders. Personalized medicine approaches-which involve parsing patients into subgroups better indicated for a particular treatment-could therefore improve patient outcomes and serve as a powerful tool in patient selection within clinical trials. Machine learning approaches can identify patient subgroups but are often not "explainable" due to the use of complex algorithms that do not mirror clinicians' natural decision-making processes. METHODS: Here we combine two analytical approaches-Personalized Advantage Index and Bayesian Rule Lists-to identify paliperidone-indicated schizophrenia patients in a way that emphasizes model explainability. We apply these approaches retrospectively to randomized, placebo-controlled clinical trial data to identify a paliperidone-indicated subgroup of schizophrenia patients who demonstrate a larger treatment effect (outcome on treatment superior than on placebo) than that of the full randomized sample as assessed with Cohen's d. For this study, the outcome corresponded to a reduction in the Positive and Negative Syndrome Scale (PANSS) total score which measures positive (e.g., hallucinations, delusions), negative (e.g., blunted affect, emotional withdrawal), and general psychopathological (e.g., disturbance of volition, uncooperativeness) symptoms in schizophrenia. RESULTS: Using our combined explainable AI approach to identify a subgroup more responsive to paliperidone than placebo, the treatment effect increased significantly over that of the full sample (p < 0.0001 for a one-sample t-test comparing the full sample Cohen's d = 0.82 and a generated distribution of subgroup Cohen's d's with mean d = 1.22, std d = 0.09). In addition, our modeling approach produces simple logical statements (if-then-else), termed a "rule list", to ease interpretability for clinicians. A majority of the rule lists generated from cross-validation found two general psychopathology symptoms, disturbance of volition and uncooperativeness, to predict membership in the paliperidone-indicated subgroup. CONCLUSIONS: These results help to technically validate our explainable AI approach to patient selection for a clinical trial by identifying a subgroup with an improved treatment effect. With these data, the explainable rule lists also suggest that paliperidone may provide an improved therapeutic benefit for the treatment of schizophrenia patients with either of the symptoms of high disturbance of volition or high uncooperativeness. TRIAL REGISTRATION: clincialtrials.gov identifier: NCT 00,083,668; prospectively registered May 28, 2004.
22A genome-by-environment interaction classifier for precision medicine: personal transcriptome response to rhinovirus identifies children prone to asthma exacerbations.PubMed
Vincent Gardeux, Joanne Berghout, Ikbel Achour, et al.
J Am Med Inform Assoc. 2017 Nov 1;24(6):1116-1126. doi: 10.1093/jamia/ocx069.
OBJECTIVE: To introduce a disease prognosis framework enabled by a robust classification scheme derived from patient-specific transcriptomic response to stimulation. MATERIALS AND METHODS: Within an illustrative case study to predict asthma exacerbation, we designed a stimulation assay that reveals individualized transcriptomic response to human rhinovirus. Gene expression from peripheral blood mononuclear cells was quantified from 23 pediatric asthmatic patients and stimulated in vitro with human rhinovirus. Responses were obtained via the single-subject gene set testing methodology "N-of-1-pathways." The classifier was trained on a related independent training dataset (n = 19). Novel visualizations of personal transcriptomic responses are provided. RESULTS: Of the 23 pediatric asthmatic patients, 12 experienced recurrent exacerbations. Our classifier, using individualized responses and trained on an independent dataset, obtained 74% accuracy (area under the receiver operating curve of 71%; 2-sided P = .039). Conventional classifiers using messenger RNA (mRNA) expression within the viral-exposed samples were unsuccessful (all patients predicted to have recurrent exacerbations; accuracy of 52%). DISCUSSION: Prognosis based on single time point, static mRNA expression alone neglects the importance of dynamic genome-by-environment interplay in phenotypic presentation. Individualized transcriptomic response quantified at the pathway (gene sets) level reveals interpretable signals related to clinical outcomes. CONCLUSION: The proposed framework provides an innovative approach to precision medicine. We show that quantifying personal pathway-level transcriptomic response to a disease-relevant environmental challenge predicts disease progression. This genome-by-environment interaction assay offers a noninvasive opportunity to translate omics data to clinical practice by improving the ability to predict disease exacerbation and increasing the potential to produce more effective treatment decisions.
23Artificial intelligence and machine learning for clinical pharmacology.PubMed
David K Ryan, Rory H Maclean, Alfred Balston, et al.
Br J Clin Pharmacol. 2024 Mar;90(3):629-639. doi: 10.1111/bcp.15930. Epub 2023 Nov 12.
Artificial intelligence (AI) will impact many aspects of clinical pharmacology, including drug discovery and development, clinical trials, personalized medicine, pharmacogenomics, pharmacovigilance and clinical toxicology. The rapid progress of AI in healthcare means clinical pharmacologists should have an understanding of AI and its implementation in clinical practice. As with any new therapy or health technology, it is imperative that AI tools are subject to robust and stringent evaluation to ensure that they enhance clinical practice in a safe and equitable manner. This review serves as an introduction to AI for the clinical pharmacologist, highlighting current applications, aspects of model development and issues surrounding evaluation and deployment. The aim of this article is to empower clinical pharmacologists to embrace and lead on the safe and effective use of AI within healthcare.
24Interpretable machine learning predicts cardiac resynchronization therapy responses from personalized biochemical and biomechanical features.PubMed
Anamul Haque, Doug Stubbs, Nina C Hubig, et al.
BMC Med Inform Decis Mak. 2022 Oct 31;22(1):282. doi: 10.1186/s12911-022-02015-0.
BACKGROUND: Cardiac Resynchronization Therapy (CRT) is a widely used, device-based therapy for patients with left ventricle (LV) failure. Unfortunately, many patients do not benefit from CRT, so there is potential value in identifying this group of non-responders before CRT implementation. Past studies suggest that predicting CRT response will require diverse variables, including demographic, biomarker, and LV function data. Accordingly, the objective of this study was to integrate diverse variable types into a machine learning algorithm for predicting individual patient responses to CRT. METHODS: We built an ensemble classification algorithm using previously acquired data from the SMART-AV CRT clinical trial (n = 794 patients). We used five-fold stratified cross-validation on 80% of the patients (n = 635) to train the model with variables collected at 0 months (before initiating CRT), and the remaining 20% of the patients (n = 159) were used as a hold-out test set for model validation. To improve model interpretability, we quantified feature importance values using SHapley Additive exPlanations (SHAP) analysis and used Local Interpretable Model-agnostic Explanations (LIME) to explain patient-specific predictions. RESULTS: Our classification algorithm incorporated 26 patient demographic and medical history variables, 12 biomarker variables, and 18 LV functional variables, which yielded correct prediction of CRT response in 71% of patients. Additional patient stratification to identify the subgroups with the highest or lowest likelihood of response showed 96% accuracy with 22 correct predictions out of 23 patients in the highest and lowest responder groups. CONCLUSION: Computationally integrating general patient characteristics, comorbidities, therapy history, circulating biomarkers, and LV function data available before CRT intervention can improve the prediction of individual patient responses.
25Artificial Intelligence and Machine Learning Approaches to Facilitate Therapeutic Drug Management and Model-Informed Precision Dosing.PubMed
Ethan A Poweleit, Alexander A Vinks, Tomoyuki Mizuno
Ther Drug Monit. 2023 Apr 1;45(2):143-150. doi: 10.1097/FTD.0000000000001078. Epub 2023 Feb 3.
BACKGROUND: Therapeutic drug monitoring (TDM) and model-informed precision dosing (MIPD) have greatly benefitted from computational and mathematical advances over the past 60 years. Furthermore, the use of artificial intelligence (AI) and machine learning (ML) approaches for supporting clinical research and support is increasing. However, AI and ML applications for precision dosing have been evaluated only recently. Given the capability of ML to handle multidimensional data, such as from electronic health records, opportunities for AI and ML applications to facilitate TDM and MIPD may be advantageous. METHODS: This review summarizes relevant AI and ML approaches to support TDM and MIPD, with a specific focus on recent applications. The opportunities and challenges associated with this integration are also discussed. RESULTS: Various AI and ML applications have been evaluated for precision dosing, including those related to concentration or exposure prediction, dose optimization, population pharmacokinetics and pharmacodynamics, quantitative systems pharmacology, and MIPD system development and support. These applications provide an opportunity for ML and pharmacometrics to operate in an integrated manner to provide clinical decision support for precision dosing. CONCLUSIONS: Although the integration of AI with precision dosing is still in its early stages and is evolving, AI and ML have the potential to work harmoniously and synergistically with pharmacometric approaches to support TDM and MIPD. Because data are increasingly shared between institutions and clinical networks and aggregated into large databases, these applications will continue to grow. The successful implementation of these approaches will depend on cross-field collaborations among clinicians and experts in informatics, ML, pharmacometrics, clinical pharmacology, and TDM.
26Model-Informed Artificial Intelligence: Reinforcement Learning for Precision Dosing.PubMed
Benjamin Ribba, Sherri Dudal, Thierry Lavé, et al.
Clin Pharmacol Ther. 2020 Apr;107(4):853-857. doi: 10.1002/cpt.1777. Epub 2020 Feb 23.
The availability of multidimensional data together with the development of modern techniques for data analysis represent an exceptional opportunity for clinical pharmacology. Data science-defined in this special issue as the novel approaches to the collection, aggregation, and analysis of data-can significantly contribute to characterize drug-response variability at the individual level, thus enabling clinical pharmacology to become a critical contributor to personalized healthcare through precision dosing. We propose a minireview of methodologies for achieving precision dosing with a focus on an artificial intelligence technique called reinforcement learning, which is currently used for individualizing dosing regimen in patients with life-threatening diseases. We highlight the interplay of such techniques with conventional pharmacokinetic/pharmacodynamic approaches and discuss applicability in drug research and early development.
27[Not Available].PubMed
Letizia Dondi, Giulia Ronconi, Leonardo Dondi, et al.
Recenti Prog Med. 2024 Dec;115(12):620-621. doi: 10.1701/4392.43928.
The role of Real-World Evidence (RWE) concerns the entire drug lifecycle; despite widely recognised in the post-marketing, it is still debated in the pre-marketing, mainly as per the identification and analysis of target populations (TPs) for new drugs or indications. Through administrative healthcare databases, Research and Health Foundation (ReS), in collaboration with experts, develops algorithms to select and analyse TPs. As of March 2024, 85 TPs in 15 clinical areas have been analysed, of which oncology is the most represented. Findings on prevalence and incidence of specific diseases (or subpopulations), patient characteristics and costs directly charged to the Italian National Health Service, are provided. These are useful for healthcare institutions and pharmaceutical companies. In the future, efforts will focus on the development of tools based on artificial intelligence and synthetic data to improve analyses of TPs and support regulatory decisions on drugs.
28Limited impact of real-world evidence in medical oncology: How can AI turn the tide.PubMed
Anna Pellat, Jakob Nikolas Kather
Eur J Cancer. 2025 Oct 1;228:115714. doi: 10.1016/j.ejca.2025.115714. Epub 2025 Aug 15.
Real-world evidence (RWE), derived from real-world health data (RWD) generated during routine clinical practice, is increasingly recognized as a valuable complement to traditional clinical trial data in medical oncology. While randomized controlled trials remain the gold standard for evidence-based medicine, they often face limitations due to rising costs, ethical constraints, and the inability to cover all treatment combinations and diverse patient populations. The current impact of RWE is limited by several challenges, including data inconsistency, quality issues, and burdensome data collection processes. Despite its role in specific areas like accelerating drug development in rare cancers, RWE has yet to achieve its full potential due to these limitations. Artificial intelligence (AI), particularly through natural language processing and large language models (LLMs), presents a transformative opportunity for RWE. AI technologies can streamline data collection, extraction, cleaning, and analysis, making the process more efficient and accurate. LLMs, for instance, can extract structured data from unstructured text and facilitate the integration of diverse data sources. This automation can significantly reduce the workload on clinicians and improve the consistency and reliability of RWD. These tools can address existing challenges and unlock the full potential of RWD, ultimately leading to improved patient outcomes and more robust oncology research.
29Video-based assessments of activities of daily living: generating real-world evidence in pediatric rare diseases.PubMed
Elisa Ferrer-Mallol, Clare Matthews, Rabia Aziza, et al.
Expert Rev Pharmacoecon Outcomes Res. 2024 Jul;24(6):713-721. doi: 10.1080/14737167.2024.2360201. Epub 2024 Jun 7.
INTRODUCTION: Preserving function and independence to perform activities of daily living (ADL) is critical for patients and carers to manage the burden of care and improve quality of life. In children living with rare diseases, video recording ADLs offer the opportunity to collect the patients' experience in a real-life setting and accurately reflect treatment effectiveness on outcomes that matter to patients and families. AREAS COVERED: We reviewed the measurement of ADL in pediatric rare diseases and the use of video to develop at-home electronic clinical outcome assessments (eCOA) by leveraging smartphone apps and artificial intelligence-based analysis. We broadly searched PubMed using Boolean combinations of the following MeSH terms 'Rare Diseases,' 'Quality of Life,' 'Activities of Daily Living,' 'Child,' 'Video Recording,' 'Outcome Assessment, Healthcare,' 'Intellectual disability,' and 'Genetic Diseases, Inborn.' Non-controlled vocabulary was used to include human pose estimation in movement analysis. EXPERT OPINION: Broad uptake of video eCOA in drug development is linked to the generation of technical and clinical validation evidence to confidently assess a patient's functional abilities. Software platforms handling video data must align with quality regulations to ensure data integrity, security, and privacy. Regulatory flexibility and optimized validation processes should facilitate video eCOA to support benefit/risk drug assessment.
30Clinical pharmacology: Current innovations and future challenges.PubMed
Jean-Luc Cracowski, Jean-Sébastien Hulot, Silvy Laporte, et al.
Fundam Clin Pharmacol. 2022 Jun;36(3):456-467. doi: 10.1111/fcp.12747. Epub 2022 Jan 26.
Clinical pharmacology is the study of drugs in humans, from first-in-human studies to randomized controlled trials (RCTs) and benefit-risk ratio assessment in large populations. The objective of this review is to present the recent innovations that may revolutionize the development of drugs in the future. On behalf of the French Society of Pharmacology and Therapeutics, we provide recommendations to address those future challenges in clinical pharmacology. Whatever the future will be, robust preliminary data on drug mechanism of action and rigorous study design will remain crucial prior to the start of pharmacological studies in human. At the present time, RCTs remain the gold standard to evaluate the efficacy of human drugs, although alternative designs (pragmatic trials, platform trials, etc.) are emerging. Innovations in healthy volunteers' studies and the contribution of new technologies such as artificial intelligence, machine learning, and internet-based trials have the potential to improve drug development. In the field of precision medicine, new disease phenotypes and endotypes will probably help to identify new pharmacological targets, responders to therapies, and patients at risk for drug adverse events. In such a moving landscape, the development of translational research through academic and private partnership, transparent sharing of clinical trial data and enhanced interactions between drug experts, patients, and the general public are priority areas for action.
31The evolution and future of integrated evidence planning.PubMed
Won Chan Lee, Chris Blanchette, Shibani Pokras, et al.
Expert Rev Pharmacoecon Outcomes Res. 2025 Jul;25(6):855-862. doi: 10.1080/14737167.2025.2497876. Epub 2025 May 5.
INTRODUCTION: Integrated Evidence Planning (IEP) is a strategic approach that optimizes drug development and market access by ensuring evidence generation aligns with regulatory, clinical, and market needs. The increasing integration of advanced technologies, including artificial intelligence/machine learning (AI/ML), natural language processing (NLP), and generative AI is set to revolutionize IEP by enhancing decision-making and improving patient access. AREAS COVERED: This article examines the role of IEP in drug development, focusing on its application across the product lifecycle, pre-clinical to post-launch. It highlights the integration of various analytical techniques, including descriptive analysis, ML, and causal inference to generate evidence. Challenges in implementing IEP, such as organizational barriers, data accessibility, and needs for specialized software tools are discussed. The evolving role of real-world evidence is emphasized, advocating for IEP as a dynamic, iterative process that adapts to market changes. Additionally, the potential of generative AI and real-time analytics to improve evidence generation and stakeholder collaboration is explored. EXPERT OPINION: The transformative potential of generative AI in IEP facilitates on-demand insights and conversational data access. However, challenges such as organizational inertia and the need for cross-functional alignment remain. Successful IEP implementation requires strong leadership, stakeholder buy-in, and optimized resource allocation to fully capitalize on its benefits.
32Integrating clinical pharmacology and artificial intelligence: potential benefits, challenges, and role of clinical pharmacologists.PubMed
Harmanjit Singh, Dwividendra Kumar Nim, Aaronbir Singh Randhawa, et al.
Expert Rev Clin Pharmacol. 2024 Apr;17(4):381-391. doi: 10.1080/17512433.2024.2317963. Epub 2024 Feb 15.
INTRODUCTION: The integration of artificial intelligence (AI) into clinical pharmacology could be a potential approach for accelerating drug discovery and development, improving patient care, and streamlining medical research processes. AREAS COVERED: We reviewed the current state of AI applications in clinical pharmacology, focusing on drug discovery and development, precision medicine, pharmacovigilance, and other ventures. Key AI applications in clinical pharmacology are examined, including machine learning, natural language processing, deep learning, and reinforcement learning etc. Additionally, the evolving role of clinical pharmacologists, ethical considerations, and challenges in implementing AI in clinical pharmacology are discussed. EXPERT OPINION: The AI could be instrumental in accelerating drug discovery, predicting drug safety and efficacy, and optimizing clinical trial designs. It can play a vital role in precision medicine by helping in personalized drug dosing, treatment selection, and predicting drug response based on genetic, clinical, and environmental factors. The role of AI in pharmacovigilance, such as signal detection and adverse event prediction, is also promising. The collaboration between clinical pharmacologists and AI experts also poses certain ethical and practical challenges. Clinical pharmacologists can be instrumental in shaping the future of AI-driven clinical pharmacology and contribute to the improvement of healthcare systems.
33Using electronic health records for clinical pharmacology research: Challenges and considerations.PubMed
Eissa Jafari, Marisa H Blackman, Jason H Karnes, et al.
Clin Transl Sci. 2024 Jul;17(7):e13871. doi: 10.1111/cts.13871.
Electronic health records (EHRs) contain a vast array of phenotypic data on large numbers of individuals, often collected over decades. Due to the wealth of information, EHR data have emerged as a powerful resource to make first discoveries and identify disparities in our healthcare system. While the number of EHR-based studies has exploded in recent years, most of these studies are directed at associations with disease rather than pharmacotherapeutic outcomes, such as drug response or adverse drug reactions. This is largely due to challenges specific to deriving drug-related phenotypes from the EHR. There is great potential for EHR-based discovery in clinical pharmacology research, and there is a critical need to address specific challenges related to accurate and reproducible derivation of drug-related phenotypes from the EHR. This review provides a detailed evaluation of challenges and considerations for deriving drug-related data from EHRs. We provide an examination of EHR-based computable phenotypes and discuss cutting-edge approaches to map medication information for clinical pharmacology research, including medication-based computable phenotypes and natural language processing. We also discuss additional considerations such as data structure, heterogeneity and missing data, rare phenotypes, and diversity within the EHR. By further understanding the complexities associated with conducting clinical pharmacology research using EHR-based data, investigators will be better equipped to design thoughtful studies with more reproducible results. Progress in utilizing EHRs for clinical pharmacology research should lead to significant advances in our ability to understand differential drug response and predict adverse drug reactions.
34Machine learning for functional protein design.PubMed
Pascal Notin, Nathan Rollins, Yarin Gal, et al.
Nat Biotechnol. 2024 Feb;42(2):216-228. doi: 10.1038/s41587-024-02127-0. Epub 2024 Feb 15.
Recent breakthroughs in AI coupled with the rapid accumulation of protein sequence and structure data have radically transformed computational protein design. New methods promise to escape the constraints of natural and laboratory evolution, accelerating the generation of proteins for applications in biotechnology and medicine. To make sense of the exploding diversity of machine learning approaches, we introduce a unifying framework that classifies models on the basis of their use of three core data modalities: sequences, structures and functional labels. We discuss the new capabilities and outstanding challenges for the practical design of enzymes, antibodies, vaccines, nanomachines and more. We then highlight trends shaping the future of this field, from large-scale assays to more robust benchmarks, multimodal foundation models, enhanced sampling strategies and laboratory automation.
35Deep learning techniques have significantly impacted protein structure prediction and protein design.PubMed
Robin Pearce, Yang Zhang
Curr Opin Struct Biol. 2021 Jun;68:194-207. doi: 10.1016/j.sbi.2021.01.007. Epub 2021 Feb 24.
Protein structure prediction and design can be regarded as two inverse processes governed by the same folding principle. Although progress remained stagnant over the past two decades, the recent application of deep neural networks to spatial constraint prediction and end-to-end model training has significantly improved the accuracy of protein structure prediction, largely solving the problem at the fold level for single-domain proteins. The field of protein design has also witnessed dramatic improvement, where noticeable examples have shown that information stored in neural-network models can be used to advance functional protein design. Thus, incorporation of deep learning techniques into different steps of protein folding and design approaches represents an exciting future direction and should continue to have a transformative impact on both fields.
36Rapid in silico directed evolution by a protein language model with EVOLVEpro.PubMed
Kaiyi Jiang, Zhaoqing Yan, Matteo Di Bernardo, et al.
Science. 2025 Jan 24;387(6732):eadr6006. doi: 10.1126/science.adr6006.
Directed protein evolution is central to biomedical applications but faces challenges such as experimental complexity, inefficient multiproperty optimization, and local maxima traps. Although in silico methods that use protein language models (PLMs) can provide modeled fitness landscape guidance, they struggle to generalize across diverse protein families and map to protein activity. We present EVOLVEpro, a few-shot active learning framework that combines PLMs and regression models to rapidly improve protein activity. EVOLVEpro surpasses current methods, yielding up to 100-fold improvements in desired properties. We demonstrate its effectiveness across six proteins in RNA production, genome editing, and antibody binding applications. These results highlight the advantages of few-shot active learning with minimal experimental data over zero-shot predictions. EVOLVEpro opens new possibilities for artificial intelligence-guided protein engineering in biology and medicine.
37De novo protein design-From new structures to programmable functions.PubMed
Tanja Kortemme
Cell. 2024 Feb 1;187(3):526-544. doi: 10.1016/j.cell.2023.12.028.
Methods from artificial intelligence (AI) trained on large datasets of sequences and structures can now "write" proteins with new shapes and molecular functions de novo, without starting from proteins found in nature. In this Perspective, I will discuss the state of the field of de novo protein design at the juncture of physics-based modeling approaches and AI. New protein folds and higher-order assemblies can be designed with considerable experimental success rates, and difficult problems requiring tunable control over protein conformations and precise shape complementarity for molecular recognition are coming into reach. Emerging approaches incorporate engineering principles-tunability, controllability, and modularity-into the design process from the beginning. Exciting frontiers lie in deconstructing cellular functions with de novo proteins and, conversely, constructing synthetic cellular signaling from the ground up. As methods improve, many more challenges are unsolved.
38Scaffolding protein functional sites using deep learning.PubMed
Jue Wang, Sidney Lisanza, David Juergens, et al.
Science. 2022 Jul 22;377(6604):387-394. doi: 10.1126/science.abn2100. Epub 2022 Jul 21.
The binding and catalytic functions of proteins are generally mediated by a small number of functional residues held in place by the overall protein structure. Here, we describe deep learning approaches for scaffolding such functional sites without needing to prespecify the fold or secondary structure of the scaffold. The first approach, "constrained hallucination," optimizes sequences such that their predicted structures contain the desired functional site. The second approach, "inpainting," starts from the functional site and fills in additional sequence and structure to create a viable protein scaffold in a single forward pass through a specifically trained RoseTTAFold network. We use these two methods to design candidate immunogens, receptor traps, metalloproteins, enzymes, and protein-binding proteins and validate the designs using a combination of in silico and experimental tests.
39Enhancing Functional Protein Design Using Heuristic Optimization and Deep Learning for Anti-Inflammatory and Gene Therapy Applications.PubMed
Ayşenur Soytürk Patat, Özkan Ufuk Nalbantoğlu
Proteins. 2025 Jul;93(7):1238-1256. doi: 10.1002/prot.26810. Epub 2025 Feb 22.
Protein sequence design is a highly challenging task, aimed at discovering new proteins that are more functional and producible under laboratory conditions than their natural counterparts. Deep learning-based approaches developed to address this problem have achieved significant success. However, these approaches often do not adequately emphasize the functional properties of proteins. In this study, we developed a heuristic optimization method to enhance key functionalities such as solubility, flexibility, and stability, while preserving the structural integrity of proteins. This method aims to reduce laboratory demands by enabling a design that is both functional and structurally sound. This approach is particularly valuable for the synthetic production of proteins with anti-inflammatory properties and those used in gene therapy. The designed proteins were initially evaluated for their ability to preserve natural structures using recovery and confidence metrics, followed by assessments with the AlphaFold tool. Additionally, natural protein sequences were mutated using a genetic algorithm and compared with those designed by our method. The results demonstrate that the protein sequences generated by our method exhibit much greater similarity to native protein sequences and structures. The code and sequences for the designed proteins are available at https://github.com/aysenursoyturk/HMHO.
40DDGemb: predicting protein stability change upon single- and multi-point variations with embeddings and deep learning.PubMed
Castrense Savojardo, Matteo Manfredi, Pier Luigi Martelli, et al.
Bioinformatics. 2024 Dec 26;41(1). doi: 10.1093/bioinformatics/btaf019.
MOTIVATION: The knowledge of protein stability upon residue variation is an important step for functional protein design and for understanding how protein variants can promote disease onset. Computational methods are important to complement experimental approaches and allow a fast screening of large datasets of variations. RESULTS: In this work, we present DDGemb, a novel method combining protein language model embeddings and transformer architectures to predict protein ΔΔG upon both single- and multi-point variations. DDGemb has been trained on a high-quality dataset derived from literature and tested on available benchmark datasets of single- and multi-point variations. DDGemb performs at the state of the art in both single- and multi-point variations. AVAILABILITY AND IMPLEMENTATION: DDGemb is available as web server at https://ddgemb.biocomp.unibo.it. Datasets used in this study are available at https://ddgemb.biocomp.unibo.it/datasets.
41Machine learning-aided engineering of hydrolases for PET depolymerization.PubMed
Hongyuan Lu, Daniel J Diaz, Natalie J Czarnecki, et al.
Nature. 2022 Apr;604(7907):662-667. doi: 10.1038/s41586-022-04599-z. Epub 2022 Apr 27.
Plastic waste poses an ecological challenge and enzymatic degradation offers one, potentially green and scalable, route for polyesters waste recycling. Poly(ethylene terephthalate) (PET) accounts for 12% of global solid waste, and a circular carbon economy for PET is theoretically attainable through rapid enzymatic depolymerization followed by repolymerization or conversion/valorization into other products. Application of PET hydrolases, however, has been hampered by their lack of robustness to pH and temperature ranges, slow reaction rates and inability to directly use untreated postconsumer plastics. Here, we use a structure-based, machine learning algorithm to engineer a robust and active PET hydrolase. Our mutant and scaffold combination (FAST-PETase: functional, active, stable and tolerant PETase) contains five mutations compared to wild-type PETase (N233K/R224Q/S121E from prediction and D186H/R280A from scaffold) and shows superior PET-hydrolytic activity relative to both wild-type and engineered alternatives between 30 and 50 °C and a range of pH levels. We demonstrate that untreated, postconsumer-PET from 51 different thermoformed products can all be almost completely degraded by FAST-PETase in 1 week. FAST-PETase can also depolymerize untreated, amorphous portions of a commercial water bottle and an entire thermally pretreated water bottle at 50 ºC. Finally, we demonstrate a closed-loop PET recycling process by using FAST-PETase and resynthesizing PET from the recovered monomers. Collectively, our results demonstrate a viable route for enzymatic plastic recycling at the industrial scale.
42ProT-VAE: Protein Transformer Variational AutoEncoder for functional protein design.PubMed
Emre Sevgen, Joshua Moller, Adrian Lange, et al.
Proc Natl Acad Sci U S A. 2025 Oct 14;122(41):e2408737122. doi: 10.1073/pnas.2408737122. Epub 2025 Oct 6.
Deep generative models have demonstrated success in learning the protein sequence to function relationship and designing synthetic sequences with engineered functionality. We introduce the Protein Transformer Variational AutoEncoder (ProT-VAE) as an accurate, generative, fast, and transferable model for data-driven protein design that blends the merits of variational autoencoders to learn interpretable, low-dimensional latent embeddings for conditional sequence design with the expressive, alignment-free featurization offered by transformer-based protein language models. We implement the model using NVIDIA's BioNeMo framework and validate its performance in retrospective functional prediction and prospective functional design. The model identifies a phenylalanine hydroxylase enzyme with 2.5× catalytic activity over wild-type, and a -carbonic anhydrase enzyme with a melting temperature elevation of ΔT = +61 C relative to the most thermostable sequence reported to date and activity in 23% v/v methyl diethanolamine at pH 11.25 and 93C corresponding to industrially relevant conditions for enzymatic carbon capture technologies. The ProT-VAE model presents a powerful and experimentally validated platform for machine learning-guided directed evolution campaigns to discover synthetic proteins with engineered function.
43AbDPP: Target-oriented antibody design with pretraining and prior biological structure knowledge.PubMed
Chenglei Yu, Xiangtian Lin, Yuxuan Cheng, et al.
Proteins. 2024 Oct;92(10):1147-1160. doi: 10.1002/prot.26676. Epub 2024 Mar 5.
Antibodies represent a crucial class of complex protein therapeutics and are essential in the treatment of a wide range of human diseases. Traditional antibody discovery methods, such as hybridoma and phage display technologies, suffer from limitations including inefficiency and a restricted exploration of the immense space of potential antibodies. To overcome these limitations, we propose a novel method for generating antibody sequences using deep learning algorithms called AbDPP (target-oriented antibody design with pretraining and prior biological knowledge). AbDPP integrates a pretrained model for antibodies with biological region information, enabling the effective use of vast antibody sequence data and intricate biological system understanding to generate sequences. To target specific antigens, AbDPP incorporates an antibody property evaluation model, which is further optimized based on evaluation results to generate more focused sequences. The efficacy of AbDPP was assessed through multiple experiments, evaluating its ability to generate amino acids, improve neutralization and binding, maintain sequence consistency, and improve sequence diversity. Results demonstrated that AbDPP outperformed other methods in terms of the performance and quality of generated sequences, showcasing its potential to enhance antibody design and screening efficiency. In summary, this study contributes to the field by offering an innovative deep learning-based method for antibody generation, addressing some limitations of traditional approaches, and underscoring the importance of integrating a specific antibody pretrained model and the biological properties of antibodies in generating novel sequences. The code and documentation underlying this article are freely available at https://github.com/zlfyj/AbDPP.
44Revolutionizing Synthetic Antibody Design: Harnessing Artificial Intelligence and Deep Sequencing Big Data for Unprecedented Advances.PubMed
Eugenio Gallo
Mol Biotechnol. 2025 Feb;67(2):410-424. doi: 10.1007/s12033-024-01064-2. Epub 2024 Feb 3.
Synthetic antibodies (Abs) represent a category of engineered proteins meticulously crafted to replicate the functions of their natural counterparts. Such Abs are generated in vitro, enabling advanced molecular alterations associated with antigen recognition, paratope site engineering, and biochemical refinements. In a parallel realm, deep sequencing has brought about a paradigm shift in molecular biology. It facilitates the prompt and cost-effective high-throughput sequencing of DNA and RNA molecules, enabling the comprehensive big data analysis of Ab transcriptomes, including specific regions of interest. Significantly, the integration of artificial intelligence (AI), based on machine- and deep- learning approaches, has fundamentally transformed our capacity to discern patterns hidden within deep sequencing big data, including distinctive Ab features and protein folding free energy landscapes. Ultimately, current AI advances can generate approximations of the most stable Ab structural configurations, enabling the prediction of de novo synthetic Abs. As a result, this manuscript comprehensively examines the latest and relevant literature concerning the intersection of deep sequencing big data and AI methodologies for the design and development of synthetic Abs. Together, these advancements have accelerated the exploration of antibody repertoires, contributing to the refinement of synthetic Ab engineering and optimizations, and facilitating advancements in the lead identification process.
45Optimization of therapeutic antibodies by predicting antigen specificity from antibody sequence via deep learning.PubMed
Derek M Mason, Simon Friedensohn, Cédric R Weber, et al.
Nat Biomed Eng. 2021 Jun;5(6):600-612. doi: 10.1038/s41551-021-00699-9. Epub 2021 Apr 15.
The optimization of therapeutic antibodies is time-intensive and resource-demanding, largely because of the low-throughput screening of full-length antibodies (approximately 1 × 10 variants) expressed in mammalian cells, which typically results in few optimized leads. Here we show that optimized antibody variants can be identified by predicting antigen specificity via deep learning from a massively diverse space of antibody sequences. To produce data for training deep neural networks, we deep-sequenced libraries of the therapeutic antibody trastuzumab (about 1 × 10 variants), expressed in a mammalian cell line through site-directed mutagenesis via CRISPR-Cas9-mediated homology-directed repair, and screened the libraries for specificity to human epidermal growth factor receptor 2 (HER2). We then used the trained neural networks to screen a computational library of approximately 1 × 10 trastuzumab variants and predict the HER2-specific subset (approximately 1 × 10 variants), which can then be filtered for viscosity, clearance, solubility and immunogenicity to generate thousands of highly optimized lead candidates. Recombinant expression and experimental testing of 30 randomly selected variants from the unfiltered library showed that all 30 retained specificity for HER2. Deep learning may facilitate antibody engineering and optimization.
46Best practices for machine learning in antibody discovery and development.PubMed
Leonard Wossnig, Norbert Furtmann, Andrew Buchanan, et al.
Drug Discov Today. 2024 Jul;29(7):104025. doi: 10.1016/j.drudis.2024.104025. Epub 2024 May 17.
In the past 40 years, therapeutic antibody discovery and development have advanced considerably, with machine learning (ML) offering a promising way to speed up the process by reducing costs and the number of experiments required. Recent progress in ML-guided antibody design and development (D&D) has been hindered by the diversity of data sets and evaluation methods, which makes it difficult to conduct comparisons and assess utility. Establishing standards and guidelines will be crucial for the wider adoption of ML and the advancement of the field. This perspective critically reviews current practices, highlights common pitfalls and proposes method development and evaluation guidelines for various ML-based techniques in therapeutic antibody D&D. Addressing challenges across the ML process, best practices are recommended for each stage to enhance reproducibility and progress.
47Navigating the landscape: A comprehensive overview of computational approaches in therapeutic antibody design and analysis.PubMed
Amar Jeet Yadav, Khushboo Bhagat, Akshit Sharma, et al.
Adv Protein Chem Struct Biol. 2025;144:33-76. doi: 10.1016/bs.apcsb.2024.10.011. Epub 2025 Jan 31.
Immunotherapy, harnessing components like antibodies, cells, and cytokines, has become a cornerstone in treating diseases such as cancer and autoimmune disorders. Therapeutic antibodies, in particular, have transformed modern medicine, providing a targeted approach that destroys disease-causing cells while sparing healthy tissues, thereby reducing the side effects commonly associated with chemotherapy. Beyond oncology, these antibodies also hold promise in addressing chronic infections where conventional therapeutics may fall short. However, antibodies identified through in vivo or in vitro methods often require extensive engineering to enhance their therapeutic potential. This optimization process, aimed at improving affinity, specificity, and reducing immunogenicity, is both challenging and costly, often involving trade-offs between critical properties. Traditional methods of antibody development, such as hybridoma technology and display techniques, are resource-intensive and time-consuming. In contrast, computational approaches offer a faster, more efficient alternative, enabling the precise design and analysis of therapeutic antibodies. These methods include sequence and structural bioinformatics approaches, next-generation sequencing-based data mining, machine learning algorithms, systems biology, immuno-informatics, and integrative approaches. These approaches are advancing the field by providing new insights and enhancing the accuracy of antibody design and analysis. In conclusion, computational approaches are essential in the development of therapeutic antibodies, significantly improving the precision and speed of discovery, optimization, and validation. Integrating these methods with experimental approaches accelerates therapeutic antibody development, paving the way for innovative strategies and treatments for various diseases ranging from cancers to autoimmune and infectious diseases.
48Rapid protein evolution by few-shot learning with a protein language model.PubMed
Kaiyi Jiang, Zhaoqing Yan, Matteo Di Bernardo, et al.
bioRxiv. 2024 Jul 18:2024.07.17.604015. doi: 10.1101/2024.07.17.604015.
Directed evolution of proteins is critical for applications in basic biological research, therapeutics, diagnostics, and sustainability. However, directed evolution methods are labor intensive, cannot efficiently optimize over multiple protein properties, and are often trapped by local maxima. -directed evolution methods incorporating protein language models (PLMs) have the potential to accelerate this engineering process, but current approaches fail to generalize across diverse protein families. We introduce EVOLVEpro, a few-shot active learning framework to rapidly improve protein activity using a combination of PLMs and protein activity predictors, achieving improved activity with as few as four rounds of evolution. EVOLVEpro substantially enhances the efficiency and effectiveness of protein evolution, surpassing current state-of-the-art methods and yielding proteins with up to 100-fold improvement of desired properties. We showcase EVOLVEpro for five proteins across three applications: T7 RNA polymerase for RNA production, a miniature CRISPR nuclease, a prime editor, and an integrase for genome editing, and a monoclonal antibody for epitope binding. These results demonstrate the advantages of few-shot active learning with small amounts of experimental data over zero-shot predictions. EVOLVEpro paves the way for broader applications of AI-guided protein engineering in biology and medicine.
49Scalable protein design using optimization in a relaxed sequence space.PubMed
Christopher Frank, Ali Khoshouei, Lara Fuβ, et al.
Science. 2024 Oct 25;386(6720):439-445. doi: 10.1126/science.adq1741. Epub 2024 Oct 24.
Machine learning (ML)-based design approaches have advanced the field of de novo protein design, with diffusion-based generative methods increasingly dominating protein design pipelines. Here, we report a "hallucination"-based protein design approach that functions in relaxed sequence space, enabling the efficient design of high-quality protein backbones over multiple scales and with broad scope of application without the need for any form of retraining. We experimentally produced and characterized more than 100 proteins. Three high-resolution crystal structures and two cryo-electron microscopy density maps of designed single-chain proteins comprising up to 1000 amino acids validate the accuracy of the method. Our pipeline can also be used to design synthetic protein-protein interactions, as validated experimentally by a set of protein heterodimers. Relaxed sequence optimization offers attractive performance with respect to designability, scope of applicability for different design problems, and scalability across protein sizes.
50A synthetic protein-level neural network in mammalian cells.PubMed
Zibo Chen, James M Linton, Shiyu Xia, et al.
Science. 2024 Dec 13;386(6727):1243-1250. doi: 10.1126/science.add8468. Epub 2024 Dec 12.
Artificial neural networks provide a powerful paradigm for nonbiological information processing. To understand whether similar principles could enable computation within living cells, we combined de novo-designed protein heterodimers and engineered viral proteases to implement a synthetic protein circuit that performs winner-take-all neural network classification. This "perceptein" circuit combines weighted input summation through reversible binding interactions with self-activation and mutual inhibition through irreversible proteolytic cleavage. These interactions collectively generate a large repertoire of distinct protein species stemming from up to eight coexpressed starting protein species. The complete system achieves multi-output signal classification with tunable decision boundaries in mammalian cells and can be used to conditionally control cell death. These results demonstrate how engineered protein-based networks can enable programmable signal classification in living cells.
51AI sheds new light on genome editing.PubMed
Zhaohui Qin, Zhaolong Deng, Chengwei Li, et al.
Trends Biotechnol. 2025 Jul 30. doi: 10.1016/j.tibtech.2025.07.011.
Artificial intelligence (AI) has revolutionized life sciences, driving transformative advances in engineering clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated (Cas)-based genome editors for therapeutic and agricultural applications. Recent breakthroughs demonstrate how deep learning accelerates the discovery, engineering, and design of next-generation genome editing tools. In this review, we explore how AI-driven approaches are supercharging genome editing in three aspects: (i) structure-based methods for discovering novel genome editors neglected by conventional methods, (ii) engineering genome editors with enhanced properties, and (iii) the de novo design of entirely new genome editors endowed with bespoke functions. Finally, we discuss the current challenges and envision the future potential of data-driven AI to unlock new possibilities in genome editing, catalyzing innovations across biology and biotechnology.
52From Code to Life: The AI-Driven Revolution in Genome Editing.PubMed
Zhidong Li, Wasi Ullah Khan, Genxiang Bai, et al.
Adv Sci (Weinh). 2025 Aug;12(30):e17029. doi: 10.1002/advs.202417029. Epub 2025 Jun 19.
Genome editing has revolutionized modern biotechnology, enabling precise modifications to DNA sequences with far-reaching applications in medicine, agriculture, and synthetic biology. Recent advancements in artificial intelligence (AI) have significantly enhanced genome editing by improving target selection, minimizing off-target effects, and optimizing CRISPR-associated systems. AI-driven models, such as deep learning-based predictors and protein language models, enable more accurate sgRNA design, novel Cas protein discovery, and enhanced gene regulatory network analysis. Additionally, AI-powered tools facilitate large-scale data integration, accelerating functional genomics and therapeutic genome editing. This review explores the intersection of AI and genome editing, highlighting key innovations, challenges, and future prospects. Despite its transformative potential, AI-driven genome editing raises ethical concerns regarding data bias, algorithmic transparency, and unintended genetic modifications. Addressing these challenges requires interdisciplinary collaboration between AI researchers, molecular biologists, and policymakers. As AI continues to evolve, its integration with genome editing will pave the way for groundbreaking advancements in precision medicine, genetic disease treatment, and sustainable agriculture.
53Precise genome editing process and its applications in plants driven by AI.PubMed
Bo Jiang, Zeyu An, Linlin Niu, et al.
Funct Integr Genomics. 2025 May 24;25(1):109. doi: 10.1007/s10142-025-01619-9.
Genome editing technologies have emerged as the keystone of biotechnological research, enabling precise gene modification. The field has evolved rapidly through revolutionary advancements, transitioning from early explorations to the breakthrough of the CRISPR-Cas system. The emergence of the CRISPR-Cas system represents a huge leap in genome editing, prompting the development of advanced tools such as base and prime editors, thereby enhancing precise genomic engineering capabilities. The rapid integration of AI across disciplines is now driving another transformative phase in genome editing, streamlining workflows and enhancing precision. The application prospects of genome editing technology are extensive, particularly in plant breeding, where it has already presented unparalleled opportunities for improving plant traits. Here, we review early genome editing technologies, including meganucleases, ZFNs, TALENs, and CRISPR-Cas systems. We also provide a detailed introduction to next-generation editing tools-such as base editors and prime editors-and their latest applications in plants. At the same time, we summarize and prospect the cutting-edge developments and future trends of genome editing technologies in combination with the rapidly rising AI technology, including optimizing editing systems, predicting the efficiency of editing sites and designing editing strategies. We are convinced that as these technologies progress and their utilization expands, they will provide pioneering solutions to global challenges, ushering in an era of health, prosperity, and sustainability.
54Predicting prime editing efficiency and product purity by deep learning.PubMed
Nicolas Mathis, Ahmed Allam, Lucas Kissling, et al.
Nat Biotechnol. 2023 Aug;41(8):1151-1159. doi: 10.1038/s41587-022-01613-7. Epub 2023 Jan 16.
Prime editing is a versatile genome editing tool but requires experimental optimization of the prime editing guide RNA (pegRNA) to achieve high editing efficiency. Here we conducted a high-throughput screen to analyze prime editing outcomes of 92,423 pegRNAs on a highly diverse set of 13,349 human pathogenic mutations that include base substitutions, insertions and deletions. Based on this dataset, we identified sequence context features that influence prime editing and trained PRIDICT (prime editing guide prediction), an attention-based bidirectional recurrent neural network. PRIDICT reliably predicts editing rates for all small-sized genetic changes with a Spearman's R of 0.85 and 0.78 for intended and unintended edits, respectively. We validated PRIDICT on endogenous editing sites as well as an external dataset and showed that pegRNAs with high (>70) versus low (<70) PRIDICT scores showed substantially increased prime editing efficiencies in different cell types in vitro (12-fold) and in hepatocytes in vivo (tenfold), highlighting the value of PRIDICT for basic and for translational research applications.
55Machine learning prediction of prime editing efficiency across diverse chromatin contexts.PubMed
Nicolas Mathis, Ahmed Allam, András Tálas, et al.
Nat Biotechnol. 2025 May;43(5):712-719. doi: 10.1038/s41587-024-02268-2. Epub 2024 Jun 21.
The success of prime editing depends on the prime editing guide RNA (pegRNA) design and target locus. Here, we developed machine learning models that reliably predict prime editing efficiency. PRIDICT2.0 assesses the performance of pegRNAs for all edit types up to 15 bp in length in mismatch repair-deficient and mismatch repair-proficient cell lines and in vivo in primary cells. With ePRIDICT, we further developed a model that quantifies how local chromatin environments impact prime editing rates.
56DeepCRISTL: deep transfer learning to predict CRISPR/Cas9 functional and endogenous on-target editing efficiency.PubMed
Shai Elkayam, Yaron Orenstein
Bioinformatics. 2022 Jun 24;38(Suppl 1):i161-i168. doi: 10.1093/bioinformatics/btac218.
MOTIVATION: CRISPR/Cas9 technology has been revolutionizing the field of gene editing in recent years. Guide RNAs (gRNAs) enable Cas9 proteins to target specific genomic loci for editing. However, editing efficiency varies between gRNAs. Thus, computational methods were developed to predict editing efficiency for any gRNA of interest. High-throughput datasets of Cas9 editing efficiencies were produced to train machine-learning models to predict editing efficiency. However, these high-throughput datasets have low correlation with functional and endogenous editing. Another difficulty arises from the fact that functional and endogenous editing efficiency is more difficult to measure, and as a result, functional and endogenous datasets are too small to train accurate machine-learning models on. RESULTS: We developed DeepCRISTL, a deep-learning model to predict the on-target efficiency given a gRNA sequence. DeepCRISTL takes advantage of high-throughput datasets to learn general patterns of gRNA on-target editing efficiency, and then uses transfer learning (TL) to fine-tune the model and fit it to the functional and endogenous prediction task. We pre-trained the DeepCRISTL model on more than 150 000 gRNAs, produced through the DeepHF study as a high-throughput dataset of three Cas9 enzymes. We improved the DeepHF model by multi-task and ensemble techniques and achieved state-of-the-art results over each of the three enzymes: up to 0.89 in Spearman correlation between predicted and measured on-target efficiencies. To fine-tune model weights to predict on-target efficiency of functional or endogenous datasets, we tested several TL approaches, with gradual learning being the overall best performer, both when pre-trained on DeepHF and when pre-trained on CRISPROn, another high-throughput dataset. DeepCRISTL outperformed state-of-the-art methods on all functional and endogenous datasets. Using saliency maps, we identified and compared the important features learned by the model in each dataset. We believe DeepCRISTL will improve prediction performance in many other CRISPR/Cas9 editing contexts by leveraging TL to utilize both high-throughput datasets, and smaller and more biologically relevant datasets, such as functional and endogenous datasets. AVAILABILITY AND IMPLEMENTATION: DeepCRISTL is available via github.com/OrensteinLab/DeepCRISTL. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.
57BoostMEC: predicting CRISPR-Cas9 cleavage efficiency through boosting models.PubMed
Oscar A Zarate, Yiben Yang, Xiaozhong Wang, et al.
BMC Bioinformatics. 2022 Oct 26;23(1):446. doi: 10.1186/s12859-022-04998-z.
BACKGROUND: In the CRISPR-Cas9 system, the efficiency of genetic modifications has been found to vary depending on the single guide RNA (sgRNA) used. A variety of sgRNA properties have been found to be predictive of CRISPR cleavage efficiency, including the position-specific sequence composition of sgRNAs, global sgRNA sequence properties, and thermodynamic features. While prevalent existing deep learning-based approaches provide competitive prediction accuracy, a more interpretable model is desirable to help understand how different features may contribute to CRISPR-Cas9 cleavage efficiency. RESULTS: We propose a gradient boosting approach, utilizing LightGBM to develop an integrated tool, BoostMEC (Boosting Model for Efficient CRISPR), for the prediction of wild-type CRISPR-Cas9 editing efficiency. We benchmark BoostMEC against 10 popular models on 13 external datasets and show its competitive performance. CONCLUSIONS: BoostMEC can provide state-of-the-art predictions of CRISPR-Cas9 cleavage efficiency for sgRNA design and selection. Relying on direct and derived sequence features of sgRNA sequences and based on conventional machine learning, BoostMEC maintains an advantage over other state-of-the-art CRISPR efficiency prediction models that are based on deep learning through its ability to produce more interpretable feature insights and predictions.
58Enhancing CRISPR-Cas9 gRNA efficiency prediction by data integration and deep learning.PubMed
Xi Xiang, Giulia I Corsi, Christian Anthon, et al.
Nat Commun. 2021 May 28;12(1):3238. doi: 10.1038/s41467-021-23576-0.
The design of CRISPR gRNAs requires accurate on-target efficiency predictions, which demand high-quality gRNA activity data and efficient modeling. To advance, we here report on the generation of on-target gRNA activity data for 10,592 SpCas9 gRNAs. Integrating these with complementary published data, we train a deep learning model, CRISPRon, on 23,902 gRNAs. Compared to existing tools, CRISPRon exhibits significantly higher prediction performances on four test datasets not overlapping with training data used for the development of these tools. Furthermore, we present an interactive gRNA design webserver based on the CRISPRon standalone software, both available via https://rth.dk/resources/crispr/ . CRISPRon advances CRISPR applications by providing more accurate gRNA efficiency predictions than the existing tools.
59Prediction of on-target and off-target activity of CRISPR-Cas13d guide RNAs using deep learning.PubMed
Hans-Hermann Wessels, Andrew Stirn, Alejandro Méndez-Mancilla, et al.
Nat Biotechnol. 2024 Apr;42(4):628-637. doi: 10.1038/s41587-023-01830-8. Epub 2023 Jul 3.
Transcriptome engineering applications in living cells with RNA-targeting CRISPR effectors depend on accurate prediction of on-target activity and off-target avoidance. Here we design and test ~200,000 RfxCas13d guide RNAs targeting essential genes in human cells with systematically designed mismatches and insertions and deletions (indels). We find that mismatches and indels have a position- and context-dependent impact on Cas13d activity, and mismatches that result in G-U wobble pairings are better tolerated than other single-base mismatches. Using this large-scale dataset, we train a convolutional neural network that we term targeted inhibition of gene expression via gRNA design (TIGER) to predict efficacy from guide sequence and context. TIGER outperforms the existing models at predicting on-target and off-target activity on our dataset and published datasets. We show that TIGER scoring combined with specific mismatches yields the first general framework to modulate transcript expression, enabling the use of RNA-targeting CRISPRs to precisely control gene dosage.
60A versatile CRISPR/Cas9 system off-target prediction tool using language model.PubMed
Weian Du, Liang Zhao, Kaichuan Diao, et al.
Commun Biol. 2025 Jun 6;8(1):882. doi: 10.1038/s42003-025-08275-6.
Genome editing with the CRISPR/Cas9 system has revolutionized life and medical sciences, particularly in treating monogenic genetic diseases by enabling long-term therapeutic effects from a single intervention. However, the CRISPR/Cas9 system can tolerate mismatches and DNA/RNA bulges at target sites, leading to unintended off-target effects that pose challenges for gene-editing therapy development. Existing high-throughput detection and in silico prediction methods are often limited to specifically designed single guide RNAs (sgRNAs) and perform poorly on unseen sequences. To address these limitations, we introduce CCLMoff, a deep learning framework for off-target prediction that incorporates a pretrained RNA language model from RNAcentral. CCLMoff captures mutual sequence information between sgRNAs and target sites and is trained on a comprehensive, updated dataset. This approach enables accurate off-target identification and strong generalization across diverse NGS-based detection datasets. Model interpretation reveals the biological importance of the seed region, underscoring CCLMoff's analytical capabilities. The development of CCLMoff lays the foundation for a comprehensive, end-to-end sgRNA design platform, enhancing both the precision and efficiency of CRISPR/Cas9-based therapeutics. CCLMoff is a versatile tool and is publicly available at github.com/duwa2/CCLMoff .
61R-CRISPR: A Deep Learning Network to Predict Off-Target Activities with Mismatch, Insertion and Deletion in CRISPR-Cas9 System.PubMed
Rui Niu, Jiajie Peng, Zhipeng Zhang, et al.
Genes (Basel). 2021 Nov 25;12(12):1878. doi: 10.3390/genes12121878.
The Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-associated protein 9 (Cas9) system is a groundbreaking gene-editing tool, which has been widely adopted in biomedical research. However, the guide RNAs in CRISPR-Cas9 system may induce unwanted off-target activities and further affect the practical application of the technique. Most existing in silico prediction methods that focused on off-target activities possess limited predictive precision and remain to be improved. Hence, it is necessary to propose a new in silico prediction method to address this problem. In this work, a deep learning framework named R-CRISPR is presented, which devises an encoding scheme to encode gRNA-target sequences into binary matrices, a convolutional neural network as feature extractor, and a recurrent neural network to predict off-target activities with mismatch, insertion, or deletion. It is demonstrated that R-CRISPR surpasses six mainstream prediction methods with a significant improvement on mismatch-only datasets verified by GUIDE-seq. Compared with the state-of-art prediction methods, R-CRISPR also achieves competitive performance on datasets with mismatch, insertion, and deletion. Furthermore, experiments show that data concatenate could influence the quality of training data, and investigate the optimal combination of datasets.
62Prediction of base editor off-targets by deep learning.PubMed
Chengdong Zhang, Yuan Yang, Tao Qi, et al.
Nat Commun. 2023 Sep 2;14(1):5358. doi: 10.1038/s41467-023-41004-3.
Due to the tolerance of mismatches between gRNA and targeting sequence, base editors frequently induce unwanted Cas9-dependent off-target mutations. Here, to develop models to predict such off-targets, we design gRNA-off- target pairs for adenine base editors (ABEs) and cytosine base editors (CBEs) and stably integrate them into the human cells. After five days of editing, we obtain valid efficiency datasets of 54,663 and 55,727 off-targets for ABEs and CBEs, respectively. We use the datasets to train deep learning models, resulting in ABEdeepoff and CBEdeepoff, which can predict off-target sites. We use these tools to predict off-targets for a panel of endogenous loci and achieve Spearman correlation values varying from 0.710 to 0.859. Finally, we develop an integrated tool that is freely accessible via an online web server http://www.deephf.com/#/bedeep/bedeepoff . These tools could facilitate minimizing the off-target effects of base editing.
63Modeling CRISPR-Cas13d on-target and off-target effects using machine learning approaches.PubMed
Xiaolong Cheng, Zexu Li, Ruocheng Shan, et al.
Nat Commun. 2023 Feb 10;14(1):752. doi: 10.1038/s41467-023-36316-3.
A major challenge in the application of the CRISPR-Cas13d system is to accurately predict its guide-dependent on-target and off-target effect. Here, we perform CRISPR-Cas13d proliferation screens and design a deep learning model, named DeepCas13, to predict the on-target activity from guide sequences and secondary structures. DeepCas13 outperforms existing methods to predict the efficiency of guides targeting both protein-coding and non-coding RNAs. Guides targeting non-essential genes display off-target viability effects, which are closely related to their on-target efficiencies. Choosing proper negative control guides during normalization mitigates the associated false positives in proliferation screens. We apply DeepCas13 to the guides targeting lncRNAs, and identify lncRNAs that affect cell viability and proliferation in multiple cell lines. The higher prediction accuracy of DeepCas13 over existing methods is extensively confirmed via a secondary CRISPR-Cas13d screen and quantitative RT-PCR experiments. DeepCas13 is freely accessible via http://deepcas13.weililab.org .
64Artificial intelligence-driven epigenetic CRISPR therapeutics: a structured multi-domain meta-analysis of therapeutic efficacy, off-target prediction, and gRNA optimization.PubMed
Mustafa Kemal Basarali, Amin Daemi, Ruhiyya Guliyeva Tahiraga, et al.
Funct Integr Genomics. 2025 Oct 25;25(1):223. doi: 10.1007/s10142-025-01725-8.
CRISPR-based epigenetic editing enables reversible regulation of gene expression without permanent DNA modification. The integration of artificial intelligence (AI) enhances guide RNA (gRNA) design, off-target prediction, and delivery optimization. We conducted a systematic review and meta-analysis (2015-2025) in accordance with PRISMA 2020 guidelines to evaluate the impact of AI on the precision, safety, and therapeutic efficacy of epigenetic CRISPR tools. From 540 screened records, 58 studies met inclusion criteria, of which 41 provided extractable quantitative data for meta-analysis and 17 contributed to qualitative synthesis. Random-effects models, subgroup analyses, and bias assessments were applied. Pooled analyses demonstrated strong positive effects across three domains: therapeutic efficacy (SMD = 1.67), gRNA optimization (SMD = 1.44), and off-target prediction (AUC = 0.79). Publication bias was minimal, and subgroup analyses indicated the strongest impact in therapeutic applications. Deep learning models were consistently associated with higher effect sizes. Qualitative synthesis revealed trends in interpretable AI, omics integration, and delivery innovations, underscoring AI's role in safer and more precise CRISPR editing. Overall, AI significantly improves the precision and therapeutic performance of CRISPR-based epigenetic tools, with the strongest effects observed in therapeutic efficacy, supporting their potential for personalized gene editing.
65Advances in viral vector-based delivery systems for gene therapy: a comprehensive review.PubMed
Kuldeep Singh, Divya Jain, Pranshul Sethi, et al.
3 Biotech. 2025 Jul;15(7):196. doi: 10.1007/s13205-025-04366-7. Epub 2025 May 30.
Gene therapy offers a revolutionary approach for treating genetic and acquired disorders by delivering therapeutic genes to target cells. Viral vectors, engineered for efficient gene delivery, have become indispensable due to their high transduction efficiency, enhanced targeting specificity, and minimized immunogenicity. This review explores recent advancements in viral vector platforms, focusing on intracellular trafficking, transgene expression, and strategies to optimize vector design. Key challenges, including immune responses, insertional mutagenesis, and manufacturing scalability, are critically examined alongside innovative solutions such as hybrid delivery systems and AI-driven capsid design. The impact of these innovations is underscored by the 28 ongoing clinical trials and 16 FDA/EMA-approved gene therapies. In addition, integrating artificial intelligence for vector optimization and CRISPR-based gene correction holds great potential for precision medicine. Future directions include refining non-viral alternatives and improving biomanufacturing for enhanced accessibility. These advancements collectively drive the field toward safer and more effective gene therapy solutions.
66Emerging Therapeutic Strategies for Heart Failure: A Comprehensive Review of Novel Pharmacological and Molecular Targets.PubMed
Elizabeth Caroline Palaparthi, Tanvi Padala, Reva Singamaneni, et al.
Cureus. 2025 Apr 1;17(4):e81573. doi: 10.7759/cureus.81573. eCollection 2025 Apr.
Heart failure (HF) is a complex clinical syndrome characterized by the heart's inability to meet the body's metabolic demands. HF remains a global health challenge with high morbidity and mortality. Outcomes of beta-blockers, angiotensin receptor-neprilysin inhibitors (ARNIs), and mineralocorticoid receptor antagonists (MRAs) in HF remain suboptimal. HF is a heterogeneous syndrome driven by neurohormonal dysregulation, fibrosis, metabolic disturbances, and inflammation, contributing to symptoms like dyspnea, fatigue, and fluid retention. Recent advances in pharmacological therapies, including sodium-glucose cotransporter 2 inhibitors (SGLT2 inhibitors), soluble guanylate cyclase stimulators (sGC stimulators), and cardiac myosin activators, have shown promise in HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF), offering mechanism-specific interventions. Moreover, molecular-targeted therapies, such as clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9) gene editing, RNA-based therapeutics, and adeno-associated virus serotype 9-sarcoplasmic reticulum calcium ATPase 2a (AAV9-SERCA2a gene) therapy, are emerging as potential disease-modifying treatments aimed at addressing genetic and inflammatory drivers of cardiomyopathies. Artificial intelligence (AI) is transforming HF care by enhancing predictive modelling, risk stratification, and precision medicine, with applications in multi-omics data integration. AI-driven tools, including machine learning (ML) algorithms, improve echocardiographic phenotyping, optimize treatment strategies, and refine patient selection for therapies. Despite these promising developments, challenges such as data quality, standardization, scalability, and regulatory barriers remain. Furthermore, gene therapies' long-term safety and efficacy are still uncertain, with concerns about immune responses, off-target effects, and sustained gene expression. Regenerative medicine strategies, including induced pluripotent stem cells (iPSC)-derived cardiomyocytes, extracellular vesicles (EVs), and 3D-bioprinted cardiac patches, offer potential solutions for myocardial repair. However, immune rejection, graft integration, and long-term viability remain significant obstacles. Additionally, high costs associated with novel biologics and gene-based therapies limit accessibility, particularly in low-resource settings. The future of HF management depends on overcoming these translational challenges. Key steps include validating AI-driven phenotyping tools in clinical trials, advancing scalable biomanufacturing technologies, and refining regulatory frameworks to facilitate clinical integration. By addressing these barriers, precision medicine, AI, and regenerative therapies can transform HF management, providing more personalized, effective, and accessible treatments and ultimately improving patient outcomes globally.
67Machine learning-based prediction models to guide the selection of Cas9 variants for efficient gene editing.PubMed
Jianbo Li, Panfeng Wu, Zhoutao Cao, et al.
Cell Rep. 2024 Feb 27;43(2):113765. doi: 10.1016/j.celrep.2024.113765. Epub 2024 Feb 14.
The increasing emergence of Cas9 variants has attracted broad interest, as these variants were designed to expand CRISPR applications. New Cas9 variants typically feature higher editing efficiency, improved editing specificity, or alternative PAM sequences. To select Cas9 variants and gRNAs for high-fidelity and efficient genome editing, it is crucial to systematically quantify the editing performances of gRNAs and develop prediction models based on high-quality datasets. Using synthetic gRNA-target paired libraries and next-generation sequencing, we compared the activity and specificity of gRNAs of four SpCas9 variants. The nucleotide composition in the PAM-distal region had more influence on the editing efficiency of HiFi Cas9 and LZ3 Cas9. We further developed machine learning models to predict the gRNA efficiency and specificity for the four Cas9 variants. To aid users from broad research areas, the machine learning models for the predictions of gRNA editing efficiency within human genome sites are available on our website.
68CRISPR enables sustainable cereal production for a greener future.PubMed
Sunny Ahmar, Babar Usman, Goetz Hensel, et al.
Trends Plant Sci. 2024 Feb;29(2):179-195. doi: 10.1016/j.tplants.2023.10.016. Epub 2023 Nov 17.
The clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) system has become the most important tool for targeted genome editing in many plant and animal species over the past decade. The CRISPR/Cas9 technology has also sparked a flood of applications and technical advancements in genome editing in the key cereal crops, including rice, wheat, maize, and barley. Here, we review advanced uses of CRISPR/Cas9 and derived systems in genome editing of cereal crops to enhance a variety of agronomically important features. We also highlight new technological advances for delivering preassembled Cas9-gRNA ribonucleoprotein (RNP)-editing systems, multiplex editing, gain-of-function strategies, the use of artificial intelligence (AI)-based tools, and combining CRISPR with novel speed breeding (SB) and vernalization strategies.
69Challenges and Opportunities in the Application of CRISPR-Cas9: A Review on Genomic Editing and Therapeutic Potentials.PubMed
Mohammad Ali Karimi, Mahdiesadat Paryan, Ghazaleh Behrouzian Fard, et al.
Med Princ Pract. 2025 Jul 17:1-17. doi: 10.1159/000547334.
Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) and the CRISPR-associated protein 9 (Cas9) constitute a revolutionary gene-editing technology, allowing precise DNA modifications with vast potential for disease treatment and the creation of genetically modified organisms. This system consists of various components designed to target specific genes, requiring efficient nuclear access within target cells through diverse delivery methods, including physical techniques and carrier-based approaches. Despite its transformative promise, CRISPR faces several challenges, including efficient cellular delivery, off-target effects, immune responses, optimizing editing efficiency, and ethical concerns. Overcoming these hurdles is crucial for fully harnessing its applications. However, CRISPR-Cas9 offers remarkable opportunities for pioneering gene therapies across various disorders, including cancer, and could revolutionize agriculture by engineering disease-resistant crops. A key strength of CRIScPR lies in its adaptability to a broad spectrum of genes, significantly enhancing its versatility. The technology's potential further expands through emerging interdisciplinary integrations, such as artificial intelligence, machine learning, and biological imaging. These advancements can refine CRISPR's precision, improve efficiency, and mitigate existing limitations, positioning it as an indispensable tool in future genetic research. Overall, CRISPR-Cas9 promises transformative progress in healthcare and agriculture, solidifying its role as a cornerstone in the evolution of genetic engineering.
70Integrating and mining diverse data in human immunological studies.PubMed
Janet C Siebert, Brandie D Wagner, Elizabeth Juarez-Colunga
Bioanalysis. 2014 Jan;6(2):209-23. doi: 10.4155/bio.13.309.
Bioanalysts and immunologists can interrogate the immune system with a variety of high-throughput technologies such as gene expression, multiplex bead arrays and flow cytometry. Conceptually, these assays support systems immunology studies, in which phenomena can be measured and correlated across biological compartments. First, however, the resulting high-dimensional data must be combined in a consistent fashion that supports analysis of the data as an integrated whole. Next, analytical methods must be applied to the hundreds or thousands of readouts. We recommend the use of a four-part analytical pipeline, consisting of data integration, hypothesis generation, prediction and hypothesis testing, and validation. We describe a variety of established methods appropriate for these integrated datasets, and highlight their application to human immunological studies. Our goal is to provide bioanalysts, immunologists and data analysts with a valuable perspective with which to approach the multiassay high-dimensional datasets generated by contemporary immunological studies.
71Identifying specificity groups in the T cell receptor repertoire.PubMed
Jacob Glanville, Huang Huang, Allison Nau, et al.
Nature. 2017 Jul 6;547(7661):94-98. doi: 10.1038/nature22976. Epub 2017 Jun 21.
T cell receptor (TCR) sequences are very diverse, with many more possible sequence combinations than T cells in any one individual. Here we define the minimal requirements for TCR antigen specificity, through an analysis of TCR sequences using a panel of peptide and major histocompatibility complex (pMHC)-tetramer-sorted cells and structural data. From this analysis we developed an algorithm that we term GLIPH (grouping of lymphocyte interactions by paratope hotspots) to cluster TCRs with a high probability of sharing specificity owing to both conserved motifs and global similarity of complementarity-determining region 3 (CDR3) sequences. We show that GLIPH can reliably group TCRs of common specificity from different donors, and that conserved CDR3 motifs help to define the TCR clusters that are often contact points with the antigenic peptides. As an independent validation, we analysed 5,711 TCRβ chain sequences from reactive CD4 T cells from 22 individuals with latent Mycobacterium tuberculosis infection. We found 141 TCR specificity groups, including 16 distinct groups containing TCRs from multiple individuals. These TCR groups typically shared HLA alleles, allowing prediction of the likely HLA restriction, and a large number of M. tuberculosis T cell epitopes enabled us to identify pMHC ligands for all five of the groups tested. Mutagenesis and de novo TCR design confirmed that the GLIPH-identified motifs were critical and sufficient for shared-antigen recognition. Thus the GLIPH algorithm can analyse large numbers of TCR sequences and define TCR specificity groups shared by TCRs and individuals, which should greatly accelerate the analysis of T cell responses and expedite the identification of specific ligands.
72Global analysis of shared T cell specificities in human non-small cell lung cancer enables HLA inference and antigen discovery.PubMed
Shin-Heng Chiou, Diane Tseng, Alexandre Reuben, et al.
Immunity. 2021 Mar 9;54(3):586-602.e8. doi: 10.1016/j.immuni.2021.02.014.
To identify disease-relevant T cell receptors (TCRs) with shared antigen specificity, we analyzed 778,938 TCRβ chain sequences from 178 non-small cell lung cancer patients using the GLIPH2 (grouping of lymphocyte interactions with paratope hotspots 2) algorithm. We identified over 66,000 shared specificity groups, of which 435 were clonally expanded and enriched in tumors compared to adjacent lung. The antigenic epitopes of one such tumor-enriched specificity group were identified using a yeast peptide-HLA A02:01 display library. These included a peptide from the epithelial protein TMEM161A, which is overexpressed in tumors and cross-reactive epitopes from Epstein-Barr virus and E. coli. Our findings suggest that this cross-reactivity may underlie the presence of virus-specific T cells in tumor infiltrates and that pathogen cross-reactivity may be a feature of multiple cancers. The approach and analytical pipelines generated in this work, as well as the specificity groups defined here, present a resource for understanding the T cell response in cancer.
73Artificial intelligence for predictive biomarker discovery in immuno-oncology: a systematic review.PubMed
A Prelaj, V Miskovic, M Zanitti, et al.
Ann Oncol. 2024 Jan;35(1):29-65. doi: 10.1016/j.annonc.2023.10.125. Epub 2023 Oct 23.
BACKGROUND: The widespread use of immune checkpoint inhibitors (ICIs) has revolutionised treatment of multiple cancer types. However, selecting patients who may benefit from ICI remains challenging. Artificial intelligence (AI) approaches allow exploitation of high-dimension oncological data in research and development of precision immuno-oncology. MATERIALS AND METHODS: We conducted a systematic literature review of peer-reviewed original articles studying the ICI efficacy prediction in cancer patients across five data modalities: genomics (including genomics, transcriptomics, and epigenomics), radiomics, digital pathology (pathomics), and real-world and multimodality data. RESULTS: A total of 90 studies were included in this systematic review, with 80% published in 2021-2022. Among them, 37 studies included genomic, 20 radiomic, 8 pathomic, 20 real-world, and 5 multimodal data. Standard machine learning (ML) methods were used in 72% of studies, deep learning (DL) methods in 22%, and both in 6%. The most frequently studied cancer type was non-small-cell lung cancer (36%), followed by melanoma (16%), while 25% included pan-cancer studies. No prospective study design incorporated AI-based methodologies from the outset; rather, all implemented AI as a post hoc analysis. Novel biomarkers for ICI in radiomics and pathomics were identified using AI approaches, and molecular biomarkers have expanded past genomics into transcriptomics and epigenomics. Finally, complex algorithms and new types of AI-based markers, such as meta-biomarkers, are emerging by integrating multimodal/multi-omics data. CONCLUSION: AI-based methods have expanded the horizon for biomarker discovery, demonstrating the power of integrating multimodal data from existing datasets to discover new meta-biomarkers. While most of the included studies showed promise for AI-based prediction of benefit from immunotherapy, none provided high-level evidence for immediate practice change. A priori planned prospective trial designs are needed to cover all lifecycle steps of these software biomarkers, from development and validation to integration into clinical practice.
74LORIS robustly predicts patient outcomes with immune checkpoint blockade therapy using common clinical, pathologic and genomic features.PubMed
Tian-Gen Chang, Yingying Cao, Hannah J Sfreddo, et al.
Nat Cancer. 2024 Aug;5(8):1158-1175. doi: 10.1038/s43018-024-00772-7. Epub 2024 Jun 3.
Despite the revolutionary impact of immune checkpoint blockade (ICB) in cancer treatment, accurately predicting patient responses remains challenging. Here, we analyzed a large dataset of 2,881 ICB-treated and 841 non-ICB-treated patients across 18 solid tumor types, encompassing a wide range of clinical, pathologic and genomic features. We developed a clinical score called LORIS (logistic regression-based immunotherapy-response score) using a six-feature logistic regression model. LORIS outperforms previous signatures in predicting ICB response and identifying responsive patients even with low tumor mutational burden or programmed cell death 1 ligand 1 expression. LORIS consistently predicts patient objective response and short-term and long-term survival across most cancer types. Moreover, LORIS showcases a near-monotonic relationship with ICB response probability and patient survival, enabling precise patient stratification. As an accurate, interpretable method using a few readily measurable features, LORIS may help improve clinical decision-making in precision medicine to maximize patient benefit. LORIS is available as an online tool at https://loris.ccr.cancer.gov/ .
75Single cell T cell landscape and T cell receptor repertoire profiling of AML in context of PD-1 blockade therapy.PubMed
Hussein A Abbas, Dapeng Hao, Katarzyna Tomczak, et al.
Nat Commun. 2021 Oct 18;12(1):6071. doi: 10.1038/s41467-021-26282-z.
In contrast to the curative effect of allogenic stem cell transplantation in acute myeloid leukemia via T cell activity, only modest responses are achieved with checkpoint-blockade therapy, which might be explained by T cell phenotypes and T cell receptor (TCR) repertoires. Here, we show by paired single-cell RNA analysis and TCR repertoire profiling of bone marrow cells in relapsed/refractory acute myeloid leukemia patients pre/post azacytidine+nivolumab treatment that the disease-related T cell subsets are highly heterogeneous, and their abundance changes following PD-1 blockade-based treatment. TCR repertoires expand and primarily emerge from CD8 cells in patients responding to treatment or having a stable disease, while TCR repertoires contract in therapyresistant patients. Trajectory analysis reveals a continuum of CD8 T cell phenotypes, characterized by differential expression of granzyme B and a bone marrow-residing memory CD8 T cell subset, in which a population with stem-like properties expressing granzyme K is enriched in responders. Chromosome 7/7q loss, on the other hand, is a cancer-intrinsic genomic marker of PD-1 blockade resistance in AML. In summary, our study reveals that adaptive T cell plasticity and genomic alterations determine responses to PD-1 blockade in acute myeloid leukemia.
76Cancer vaccines and the future of immunotherapy.PubMed
Orrin Pail, Matthew J Lin, Theodora Anagnostou, et al.
Lancet. 2025 Jul 12;406(10499):189-202. doi: 10.1016/S0140-6736(25)00553-7. Epub 2025 Jun 18.
Vaccines have had a major impact on the control of infectious disease, most recently by helping to combat the COVID-19 pandemic. Prophylactic cancer vaccines have prevented several malignancies by protecting against cancer-causing pathogens. By contrast, therapeutic vaccines training the immune system to eliminate established tumours are now showing real promise in clinical settings. In the adjuvant setting, vaccines against melanoma and pancreatic cancer appear to be reducing minimal residual disease and relapse. In the macrometastatic setting, in-situ vaccines have induced systemic regressions in advanced-stage lung and breast cancers and lymphomas. More effective cancer vaccines are being developed through having a deeper understanding of crucial cellular factors in tumour immunology, the incorporation of newer vaccine components to effectively mobilise and activate cells, the use of omics and artificial intelligence in vaccine design, and addition of immune checkpoint blockade. In this Viewpoint, we analyse cancer vaccine trials, the strengths and limitations of different vaccine approaches, and we discuss how the next generation of cancer vaccines can help improve patient outcomes and quality of life.
77Artificial Intelligence for Vaccine Design.PubMed
Peter McCaffrey
Methods Mol Biol. 2022;2412:3-13. doi: 10.1007/978-1-0716-1892-9_1.
Often likened to "the new electricity," artificial intelligence (AI) has broad and sweeping impact in many areas. Perhaps most exciting among these are in bioinformatics as AI allows for new and increasingly powerful ways of understanding genomics, proteomics, and immunology, just to name a few areas. Also exciting is a parallel growth in high-throughput assays including sequencing which will further accelerate the development and use of AI in biomedicine. In this chapter, we will discuss artificial intelligence and deep leaning in particular, and we will review how such approaches are enhancing and even reshaping vaccine design in terms of epitope detection and optimization. Moreover, we discuss how AI is particularly valuable to the design of mRNA vaccines including in research and production. Finally, we will discuss several additional areas across trials and operations where AI will have pervasive impact on the development of vaccines going forward.
78Tumor neoantigens: Novel strategies for application of cancer immunotherapy.PubMed
Hanyang Guan, Yue Wu, L U Li, et al.
Oncol Res. 2023 Jun 27;31(4):437-448. doi: 10.32604/or.2023.029924. eCollection 2023.
Neoantigen-targeted immunotherapy is a rapidly advancing field that holds great promise for treating cancer. The recognition of antigens by immune cells is a crucial step in tumor-specific killing, and neoantigens generated by mutations in cancer cells possess high immunogenicity and are selectively expressed in tumor cells, making them an attractive therapeutic target. Currently, neoantigens find utility in various domains, primarily in the realm of neoantigen vaccines such as DC vaccines, nucleic acid vaccines, and synthetic long peptide vaccines. Additionally, they hold promise in adoptive cell therapy, encompassing tumor-infiltrating cells, T cell receptors, and chimeric antigen receptors which are expressed by genetically modified T cells. In this review, we summarized recent progress in the clinical use of tumor vaccines and adoptive cell therapy targeting neoantigens, discussed the potential of neoantigen burden as an immune checkpoint in clinical settings. With the aid of state-of-the-art sequencing and bioinformatics technologies, together with significant advancements in artificial intelligence, we anticipated that neoantigens will be fully exploited for personalized tumor immunotherapy, from screening to clinical application.
79Personalized cancer vaccine design using AI-powered technologies.PubMed
Anant Kumar, Shriniket Dixit, Kathiravan Srinivasan, et al.
Front Immunol. 2024 Nov 8;15:1357217. doi: 10.3389/fimmu.2024.1357217. eCollection 2024.
Immunotherapy has ushered in a new era of cancer treatment, yet cancer remains a leading cause of global mortality. Among various therapeutic strategies, cancer vaccines have shown promise by activating the immune system to specifically target cancer cells. While current cancer vaccines are primarily prophylactic, advancements in targeting tumor-associated antigens (TAAs) and neoantigens have paved the way for therapeutic vaccines. The integration of artificial intelligence (AI) into cancer vaccine development is revolutionizing the field by enhancing various aspect of design and delivery. This review explores how AI facilitates precise epitope design, optimizes mRNA and DNA vaccine instructions, and enables personalized vaccine strategies by predicting patient responses. By utilizing AI technologies, researchers can navigate complex biological datasets and uncover novel therapeutic targets, thereby improving the precision and efficacy of cancer vaccines. Despite the promise of AI-powered cancer vaccines, significant challenges remain, such as tumor heterogeneity and genetic variability, which can limit the effectiveness of neoantigen prediction. Moreover, ethical and regulatory concerns surrounding data privacy and algorithmic bias must be addressed to ensure responsible AI deployment. The future of cancer vaccine development lies in the seamless integration of AI to create personalized immunotherapies that offer targeted and effective cancer treatments. This review underscores the importance of interdisciplinary collaboration and innovation in overcoming these challenges and advancing cancer vaccine development.
80Neoantigen-driven personalized tumor therapy: An update from discovery to clinical application.PubMed
Na Xie, Guobo Shen, Canhua Huang, et al.
Chin Med J (Engl). 2025 Sep 5;138(17):2057-2090. doi: 10.1097/CM9.0000000000003708. Epub 2025 Aug 4.
Neoantigens exhibit high immunogenic potential and confer a uniqueness to tumor cells, making them ideal targets for personalized cancer immunotherapy. Neoantigens originate from tumor-specific genetic alterations, abnormal viral infections, or other biological mechanisms, including atypical RNA splicing events and post-translational modifications (PTMs). These neoantigens are recognized as foreign by the immune system, eliciting an immune response that largely bypasses conventional mechanisms of central and peripheral tolerance. Advances in next-generation sequencing (NGS), mass spectrometry (MS), and artificial intelligence (AI) have greatly expedited the rapid detection and forecasting of neoantigens, markedly propelling the development of diverse immunotherapeutic strategies, including cancer vaccines, adoptive cell therapy, and antibody treatment. In this review, we comprehensively explore the discovery and characterization of neoantigens and their clinical use within promising immunotherapeutic frameworks. Additionally, we address the current landscape of neoantigen research, the intrinsic challenges of the field, and potential pathways for clinical application in cancer treatment.
81Systems Immunology: Revealing Influenza Immunological Imprint.PubMed
Adriana Tomic, Andrew J Pollard, Mark M Davis
Viruses. 2021 May 20;13(5):948. doi: 10.3390/v13050948.
Understanding protective influenza immunity and identifying immune correlates of protection poses a major challenge and requires an appreciation of the immune system in all of its complexity. While adaptive immune responses such as neutralizing antibodies and influenza-specific T lymphocytes are contributing to the control of influenza virus, key factors of long-term protection are not well defined. Using systems immunology, an approach that combines experimental and computational methods, we can capture the systems-level state of protective immunity and reveal the essential pathways that are involved. New approaches and technological developments in systems immunology offer an opportunity to examine roles and interrelationships of clinical, biological, and genetic factors in the control of influenza infection and have the potential to lead to novel discoveries about influenza immunity that are essential for the development of more effective vaccines to prevent future pandemics. Here, we review recent developments in systems immunology that help to reveal key factors mediating protective immunity.
82The artificial intelligence and machine learning in lung cancer immunotherapy.PubMed
Qing Gao, Luyu Yang, Mingjun Lu, et al.
J Hematol Oncol. 2023 May 24;16(1):55. doi: 10.1186/s13045-023-01456-y.
Since the past decades, more lung cancer patients have been experiencing lasting benefits from immunotherapy. It is imperative to accurately and intelligently select appropriate patients for immunotherapy or predict the immunotherapy efficacy. In recent years, machine learning (ML)-based artificial intelligence (AI) was developed in the area of medical-industrial convergence. AI can help model and predict medical information. A growing number of studies have combined radiology, pathology, genomics, proteomics data in order to predict the expression levels of programmed death-ligand 1 (PD-L1), tumor mutation burden (TMB) and tumor microenvironment (TME) in cancer patients or predict the likelihood of immunotherapy benefits and side effects. Finally, with the advancement of AI and ML, it is believed that "digital biopsy" can replace the traditional single assessment method to benefit more cancer patients and help clinical decision-making in the future. In this review, the applications of AI in PD-L1/TMB prediction, TME prediction and lung cancer immunotherapy are discussed.
83Biology-guided deep learning predicts prognosis and cancer immunotherapy response.PubMed
Yuming Jiang, Zhicheng Zhang, Wei Wang, et al.
Nat Commun. 2023 Aug 23;14(1):5135. doi: 10.1038/s41467-023-40890-x.
Substantial progress has been made in using deep learning for cancer detection and diagnosis in medical images. Yet, there is limited success on prediction of treatment response and outcomes, which has important implications for personalized treatment strategies. A significant hurdle for clinical translation of current data-driven deep learning models is lack of interpretability, often attributable to a disconnect from the underlying pathobiology. Here, we present a biology-guided deep learning approach that enables simultaneous prediction of the tumor immune and stromal microenvironment status as well as treatment outcomes from medical images. We validate the model for predicting prognosis of gastric cancer and the benefit from adjuvant chemotherapy in a multi-center international study. Further, the model predicts response to immune checkpoint inhibitors and complements clinically approved biomarkers. Importantly, our model identifies a subset of mismatch repair-deficient tumors that are non-responsive to immunotherapy and may inform the selection of patients for combination treatments.
84Antimicrobial resistance: a concise update.PubMed
Charlotte S Ho, Carlos T H Wong, Thet Tun Aung, et al.
Lancet Microbe. 2025 Jan;6(1):100947. doi: 10.1016/j.lanmic.2024.07.010. Epub 2024 Sep 18.
Antimicrobial resistance (AMR) is a serious threat to global public health, with approximately 5 million deaths associated with bacterial AMR in 2019. Tackling AMR requires a multifaceted and cohesive approach that ranges from increased understanding of mechanisms and drivers at the individual and population levels, AMR surveillance, antimicrobial stewardship, improved infection prevention and control measures, and strengthened global policies and funding to development of novel antimicrobial therapeutic strategies. In this rapidly advancing field, this Review provides a concise update on AMR, encompassing epidemiology, evolution, underlying mechanisms (primarily those related to last-line or newer generation of antibiotics), infection prevention and control measures, access to antibiotics, antimicrobial stewardship, AMR surveillance, and emerging non-antibiotic therapeutic approaches. The Review also discusses the potential roles of artificial intelligence in addressing AMR, including antimicrobial susceptibility testing, AMR surveillance, antimicrobial stewardship, diagnosis, and antimicrobial drug discovery and development. This Review highlights the urgent need for addressing the global effects of AMR and for rapid advancement of relevant technology in this dynamic field.
85From Data to Decisions: Leveraging Artificial Intelligence and Machine Learning in Combating Antimicrobial Resistance - a Comprehensive Review.PubMed
José M Pérez de la Lastra, Samuel J T Wardell, Tarun Pal, et al.
J Med Syst. 2024 Aug 1;48(1):71. doi: 10.1007/s10916-024-02089-5.
The emergence of drug-resistant bacteria poses a significant challenge to modern medicine. In response, Artificial Intelligence (AI) and Machine Learning (ML) algorithms have emerged as powerful tools for combating antimicrobial resistance (AMR). This review aims to explore the role of AI/ML in AMR management, with a focus on identifying pathogens, understanding resistance patterns, predicting treatment outcomes, and discovering new antibiotic agents. Recent advancements in AI/ML have enabled the efficient analysis of large datasets, facilitating the reliable prediction of AMR trends and treatment responses with minimal human intervention. ML algorithms can analyze genomic data to identify genetic markers associated with antibiotic resistance, enabling the development of targeted treatment strategies. Additionally, AI/ML techniques show promise in optimizing drug administration and developing alternatives to traditional antibiotics. By analyzing patient data and clinical outcomes, these technologies can assist healthcare providers in diagnosing infections, evaluating their severity, and selecting appropriate antimicrobial therapies. While integration of AI/ML in clinical settings is still in its infancy, advancements in data quality and algorithm development suggest that widespread clinical adoption is forthcoming. In conclusion, AI/ML holds significant promise for improving AMR management and treatment outcome.
86The Role of Artificial Intelligence and Machine Learning Models in Antimicrobial Stewardship in Public Health: A Narrative Review.PubMed
Flavia Pennisi, Antonio Pinto, Giovanni Emanuele Ricciardi, et al.
Antibiotics (Basel). 2025 Jan 30;14(2):134. doi: 10.3390/antibiotics14020134.
Antimicrobial resistance (AMR) poses a critical global health threat, necessitating innovative approaches in antimicrobial stewardship (AMS). Artificial intelligence (AI) and machine learning (ML) have emerged as transformative tools in this domain, enabling data-driven interventions to optimize antibiotic use and combat resistance. This comprehensive review explores the multifaceted role of AI and ML models in enhancing antimicrobial stewardship efforts across healthcare systems. AI-powered predictive analytics can identify patterns of resistance, forecast outbreaks, and guide personalized antibiotic therapies by leveraging large-scale clinical and epidemiological data. ML algorithms facilitate rapid pathogen identification, resistance profiling, and real-time monitoring, enabling precise decision making. These technologies also support the development of advanced diagnostic tools, reducing the reliance on broad-spectrum antibiotics and fostering timely, targeted treatments. In public health, AI-driven surveillance systems improve the detection of AMR trends and enhance global monitoring capabilities. By integrating diverse data sources-such as electronic health records, laboratory results, and environmental data-ML models provide actionable insights to policymakers, healthcare providers, and public health officials. Additionally, AI applications in antimicrobial stewardship programs (ASPs) promote adherence to prescribing guidelines, evaluate intervention outcomes, and optimize resource allocation. Despite these advancements, challenges such as data quality, algorithm transparency, and ethical considerations must be addressed to maximize the potential of AI and ML in this field. Future research should focus on developing interpretable models and fostering interdisciplinary collaborations to ensure the equitable and sustainable integration of AI into antimicrobial stewardship initiatives.
87Artificial intelligence and infectious disease diagnostics: state of the art and future perspectives.PubMed
Luca Miglietta, Timothy M Rawson, Ronald Galiwango, et al.
Lancet Infect Dis. 2025 Sep 16. doi: 10.1016/S1473-3099(25)00354-8.
Artificial intelligence (AI) is reshaping infectious disease diagnostics by supporting clinical decision making, optimising laboratory and clinical workflows, and enabling real-time disease surveillance. AI approaches improve pathogen detection, antimicrobial stewardship, and treatment monitoring, enhancing diagnostic accuracy, efficiency, and scalability. The role of AI in combating antimicrobial resistance is particularly significant, enabling rapid pathogen identification and personalised treatment. Despite progress over the past two decades, widespread AI adoption in infectious disease diagnostics faces challenges. In high-income countries, fragmented data ecosystems, incomplete datasets, and algorithmic bias hinder clinical integration. Meanwhile, low-income and middle-income countries contend with limited digital infrastructure, unstandardised data, and financial constraints, exacerbating disparities in diagnostic access. Further barriers include concerns over interoperability, data privacy, cybersecurity, and the regulation of AI implementation. This paper examines the role of AI in infectious disease diagnostics, highlighting both opportunities and limitations. It underscores the need for coordinated investments in digital infrastructure, harmonised data-sharing frameworks, and clinician engagement to support equitable, sustainable adoption. Addressing these challenges will enable health-care systems to harness the potential of AI to improve infectious disease detection, prevention, and management of infectious diseases, thereby strengthening global health resilience.
88Emerging Microorganisms and Infectious Diseases: One Health Approach for Health Shared Vision.PubMed
Maria Vittoria Ristori, Valerio Guarrasi, Paolo Soda, et al.
Genes (Basel). 2024 Jul 11;15(7):908. doi: 10.3390/genes15070908.
Emerging infectious diseases (EIDs) are newly emerging and reemerging infectious diseases. The National Institute of Allergy and Infectious Diseases identifies the following as emerging infectious diseases: SARS, MERS, COVID-19, influenza, fungal diseases, plague, schistosomiasis, smallpox, tick-borne diseases, and West Nile fever. The factors that should be taken into consideration are the genetic adaptation of microbial agents and the characteristics of the human host or environment. The new approach to identifying new possible pathogens will have to go through the One Health approach and omics integration data, which are capable of identifying high-priority microorganisms in a short period of time. New bioinformatics technologies enable global integration and sharing of surveillance data for rapid public health decision-making to detect and prevent epidemics and pandemics, ensuring timely response and effective prevention measures. Machine learning tools are being more frequently utilized in the realm of infectious diseases to predict sepsis in patients, diagnose infectious diseases early, and forecast the effectiveness of treatment or the appropriate choice of antibiotic regimen based on clinical data. We will discuss emerging microorganisms, omics techniques applied to infectious diseases, new computational solutions to evaluate biomarkers, and innovative tools that are useful for integrating omics data and electronic medical records data for the clinical management of emerging infectious diseases.
89Identification of genetic relationships and subspecies signatures in Xylella fastidiosa.PubMed
Nicolas Denancé, Martial Briand, Romain Gaborieau, et al.
BMC Genomics. 2019 Mar 25;20(1):239. doi: 10.1186/s12864-019-5565-9.
BACKGROUND: The phytopathogenic bacterium Xylella fastidiosa was thought to be restricted to the Americas where it infects and kills numerous hosts. Its detection worldwide has been blooming since 2013 in Europe and Asia. Genetically diverse, this species is divided into six subspecies but genetic traits governing this classification are poorly understood. RESULTS: SkIf (Specific k-mers Identification) was designed and exploited for comparative genomics on a dataset of 46 X. fastidiosa genomes, including seven newly sequenced individuals. It was helpful to quickly check the synonymy between strains from different collections. SkIf identified specific SNPs within 16S rRNA sequences that can be employed for predicting the distribution of Xylella through data mining. Applied to inter- and intra-subspecies analyses, it identified specific k-mers in genes affiliated to differential gene ontologies. Chemotaxis-related genes more prevalently possess specific k-mers in genomes from subspecies fastidiosa, morus and sandyi taken as a whole group. In the subspecies pauca increased abundance of specific k-mers was found in genes associated with the bacterial cell wall/envelope/plasma membrane. Most often, the k-mer specificity occurred in core genes with non-synonymous SNPs in their sequences in genomes of the other subspecies, suggesting putative impact in the protein functions. The presence of two integrative and conjugative elements (ICEs) was identified, one chromosomic and an entire plasmid in a single strain of X. fastidiosa subsp. pauca. Finally, a revised taxonomy of X. fastidiosa into three major clades defined by the subspecies pauca (clade I), multiplex (clade II) and the combination of fastidiosa, morus and sandyi (clade III) was strongly supported by k-mers specifically associated with these subspecies. CONCLUSIONS: SkIf is a robust and rapid software, freely available, that can be dedicated to the comparison of sequence datasets and is applicable to any field of research. Applied to X. fastidiosa, an emerging pathogen in Europe, it provided an important resource to mine for identifying genetic markers of subspecies to optimize the strategies attempted to limit the pathogen dissemination in novel areas.
90Artificial intelligence-clinical decision support system for enhanced infectious disease management: Accelerating ceftazidime-avibactam resistance detection in Klebsiella pneumoniae.PubMed
Tai-Han Lin, Hsing-Yi Chung, Ming-Jr Jian, et al.
J Infect Public Health. 2024 Oct;17(10):102541. doi: 10.1016/j.jiph.2024.102541. Epub 2024 Sep 10.
BACKGROUND: Effective and rapid diagnostic strategies are required to manage antibiotic resistance in Klebsiella pneumonia (KP). This study aimed to design an artificial intelligence-clinical decision support system (AI-CDSS) using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and machine learning for the rapid detection of ceftazidime-avibactam (CZA) resistance in KP to improve clinical decision-making processes. METHODS: Out of 107,721 bacterial samples, 675 specimens of KP with suspected multi-drug resistance were selected. These specimens were collected from a tertiary hospital and four secondary hospitals between 2022 and 2023 to evaluate CZA resistance. We used MALDI-TOF MS and machine learning to develop an AI-CDSS with enhanced speed of resistance detection. RESULTS: Machine learning models, especially light gradient boosting machines (LGBM), exhibited an area under the curve (AUC) of 0.95, indicating high accuracy. The predictive models formed the core of our newly developed AI-CDSS, enabling clinical decisions quicker than traditional methods using culture and antibiotic susceptibility testing by a day. CONCLUSIONS: The study confirms that MALDI-TOF MS, integrated with machine learning, can swiftly detect CZA resistance. Incorporating this insight into an AI-CDSS could transform clinical workflows, giving healthcare professionals immediate, crucial insights for shaping treatment plans. This approach promises to be a template for future anti-resistance strategies, emphasizing the vital importance of advanced diagnostics in enhancing public health outcomes.
91ESCMID workshop: Artificial intelligence and machine learning in medical microbiology diagnostics.PubMed
Mariella Greutmann, Karsten Borgwardt, Sarah Brüningk, et al.
Microbes Infect. 2025 Sep 5:105562. doi: 10.1016/j.micinf.2025.105562.
Rapid advancements in artificial intelligence (AI) and machine learning (ML) offer significant potential to transform medical microbiology diagnostics, improving pathogen identification, antimicrobial susceptibility prediction and outbreak detection. To address these opportunities and challenges, the ESCMID workshop, "Artificial Intelligence and Machine Learning in Medical Microbiology Diagnostics", was held in Zurich, Switzerland, from June 2-5, 2025. The course featured expert lectures, practical sessions and panel discussions covering foundational ML concepts and deep learning architectures, data interoperability, quality control processes, model development and validation strategies. Key applications discussed included whole-genome sequencing for antimicrobial resistance detection, AI-enhanced digital microscopy automation and MALDI-TOF mass spectrometry-based diagnostics. Participants gained hands-on experience with essential AI tools and platforms. Special emphasis was placed on standardised laboratory protocols, regulatory compliance and ethical considerations, including data governance and patient privacy. Panel sessions further highlighted critical issues of equity, global disparities in AI access, sustainability and environmental impacts related to AI infrastructure. The workshop concluded by underscoring a necessity for ongoing interdisciplinary collaboration, continued education, and substantial investment in equitable AI infrastructure to realise the full potential of AI in clinical diagnostics.
92Application of machine learning based genome sequence analysis in pathogen identification.PubMed
Yunqiu Gao, Min Liu
Front Microbiol. 2024 Oct 2;15:1474078. doi: 10.3389/fmicb.2024.1474078. eCollection 2024.
Infectious diseases caused by pathogenic microorganisms pose a serious threat to human health. Despite advances in molecular biology, genetics, computation, and medicinal chemistry, infectious diseases remain a significant public health concern. Addressing the challenges posed by pathogen outbreaks, pandemics, and antimicrobial resistance requires concerted interdisciplinary efforts. With the development of computer technology and the continuous exploration of artificial intelligence(AI)applications in the biomedical field, the automatic morphological recognition and image processing of microbial images under microscopes have advanced rapidly. The research team of Institute of Microbiology, Chinese Academy of Sciences has developed a single cell microbial identification technology combining Raman spectroscopy and artificial intelligence. Through laser Raman acquisition system and convolutional neural network analysis, the average accuracy rate of 95.64% has been achieved, and the identification can be completed in only 5 min. These technologies have shown substantial advantages in the visible morphological detection of pathogenic microorganisms, expanding anti-infective drug discovery, enhancing our understanding of infection biology, and accelerating the development of diagnostics. In this review, we discuss the application of AI-based machine learning in image analysis, genome sequencing data analysis, and natural language processing (NLP) for pathogen identification, highlighting the significant role of artificial intelligence in pathogen diagnosis. AI can improve the accuracy and efficiency of diagnosis, promote early detection and personalized treatment, and enhance public health safety.
93Antimicrobial resistance crisis: could artificial intelligence be the solution?PubMed
Guang-Yu Liu, Dan Yu, Mei-Mei Fan, et al.
Mil Med Res. 2024 Jan 23;11(1):7. doi: 10.1186/s40779-024-00510-1.
Antimicrobial resistance is a global public health threat, and the World Health Organization (WHO) has announced a priority list of the most threatening pathogens against which novel antibiotics need to be developed. The discovery and introduction of novel antibiotics are time-consuming and expensive. According to WHO's report of antibacterial agents in clinical development, only 18 novel antibiotics have been approved since 2014. Therefore, novel antibiotics are critically needed. Artificial intelligence (AI) has been rapidly applied to drug development since its recent technical breakthrough and has dramatically improved the efficiency of the discovery of novel antibiotics. Here, we first summarized recently marketed novel antibiotics, and antibiotic candidates in clinical development. In addition, we systematically reviewed the involvement of AI in antibacterial drug development and utilization, including small molecules, antimicrobial peptides, phage therapy, essential oils, as well as resistance mechanism prediction, and antibiotic stewardship.
94De novo design of self-assembling peptides with antimicrobial activity guided by deep learning.PubMed
Huayang Liu, Zilin Song, Yu Zhang, et al.
Nat Mater. 2025 Aug;24(8):1295-1306. doi: 10.1038/s41563-025-02164-3. Epub 2025 Mar 14.
Bioinspired materials based on self-assembling peptides are promising for tackling various challenges in biomedical engineering. While contemporary data-driven approaches have led to the discovery of self-assembling peptides with various structures and properties, predicting the functionalities of these materials is still challenging. Here we describe the deep learning-guided de novo design of antimicrobial materials based on self-assembling peptides targeting bacterial membranes to address the emerging problem of bacterial drug resistance. Our approach integrates non-natural amino acids for enhanced peptide self-assembly and effectively predicts the functional activity of the self-assembling peptide materials with minimal experimental annotation. The designed self-assembling peptide leader displays excellent in vivo therapeutic efficacy against intestinal bacterial infection in mice. Moreover, it exhibits an enhanced biofilm eradication capability and does not induce acquired drug resistance. Mechanistic studies reveal that the designed peptide can self-assemble on bacterial membranes to form nanofibrous structures for killing multidrug-resistant bacteria. This work thus provides a strategy to discover functional peptide materials by customized design.
95Targeted genome mining for microbial antitumor agents acting through DNA intercalation.PubMed
Zhijie Zhao, Guiyun Zhao, Yi Chai, et al.
Synth Syst Biotechnol. 2023 Jul 22;8(3):520-526. doi: 10.1016/j.synbio.2023.07.003. eCollection 2023 Sep.
Microbial natural products have been one of the most important sources for drug development. In the current postgenomic era, sequence-driven approaches for natural product discovery are becoming increasingly popular. Here, we develop an effective genome mining strategy for the targeted discovery of microbial metabolites with antitumor activities. Our method employs -like genes as genetic markers, which have been identified in the biosynthetic gene clusters (BGCs) of several chemotherapeutic drugs of microbial origin and confer self-resistance to the corresponding producers. Through systematic genomic analysis of gifted actinobacteria genera, identification of -like gene-containing BGCs, and targeted isolation of products from a BGC prioritized for metabolic analysis, we identified a new tetracycline-type DNA intercalator timmycins. Our results thus provide a new genome mining strategy for the efficient discovery of antitumor agents acting through DNA intercalation.
96Mining biology for antibiotic discovery.PubMed
Cesar de la Fuente-Nunez
PLoS Biol. 2024 Nov 26;22(11):e3002946. doi: 10.1371/journal.pbio.3002946. eCollection 2024 Nov.
The rise of antibiotic resistance calls for innovative solutions. The realization that biology can be mined digitally using artificial intelligence has revealed a new paradigm for antibiotic discovery, offering hope in the fight against superbugs.
97Artificial intelligence for modelling infectious disease epidemics.PubMed
Moritz U G Kraemer, Joseph L-H Tsui, Serina Y Chang, et al.
Nature. 2025 Feb;638(8051):623-635. doi: 10.1038/s41586-024-08564-w. Epub 2025 Feb 19.
Infectious disease threats to individual and public health are numerous, varied and frequently unexpected. Artificial intelligence (AI) and related technologies, which are already supporting human decision making in economics, medicine and social science, have the potential to transform the scope and power of infectious disease epidemiology. Here we consider the application to infectious disease modelling of AI systems that combine machine learning, computational statistics, information retrieval and data science. We first outline how recent advances in AI can accelerate breakthroughs in answering key epidemiological questions and we discuss specific AI methods that can be applied to routinely collected infectious disease surveillance data. Second, we elaborate on the social context of AI for infectious disease epidemiology, including issues such as explainability, safety, accountability and ethics. Finally, we summarize some limitations of AI applications in this field and provide recommendations for how infectious disease epidemiology can harness most effectively current and future developments in AI.
98Development of a Bioinformatics Framework for Identification and Validation of Genomic Biomarkers and Key Immunopathology Processes and Controllers in Infectious and Non-infectious Severe Inflammatory Response Syndrome.PubMed
Dong Ling Tong, Karen E Kempsell, Tamas Szakmany, et al.
Front Immunol. 2020 Mar 31;11:380. doi: 10.3389/fimmu.2020.00380. eCollection 2020.
Sepsis is defined as dysregulated host response caused by systemic infection, leading to organ failure. It is a life-threatening condition, often requiring admission to an intensive care unit (ICU). The causative agents and processes involved are multifactorial but are characterized by an overarching inflammatory response, sharing elements in common with severe inflammatory response syndrome (SIRS) of non-infectious origin. Sepsis presents with a range of pathophysiological and genetic features which make clinical differentiation from SIRS very challenging. This may reflect a poor understanding of the key gene inter-activities and/or pathway associations underlying these disease processes. Improved understanding is critical for early differential recognition of sepsis and SIRS and to improve patient management and clinical outcomes. Judicious selection of gene biomarkers suitable for development of diagnostic tests/testing could make differentiation of sepsis and SIRS feasible. Here we describe a methodologic framework for the identification and validation of biomarkers in SIRS, sepsis and septic shock patients, using a 2-tier gene screening, artificial neural network (ANN) data mining technique, using previously published gene expression datasets. Eight key hub markers have been identified which may delineate distinct, core disease processes and which show potential for informing underlying immunological and pathological processes and thus patient stratification and treatment. These do not show sufficient fold change differences between the different disease states to be useful as primary diagnostic biomarkers, but are instrumental in identifying candidate pathways and other associated biomarkers for further exploration.
99Artificial intelligence and its application in clinical microbiology.PubMed
Assia Mairi, Lamia Hamza, Abdelaziz Touati
Expert Rev Anti Infect Ther. 2025 Jul;23(7):469-490. doi: 10.1080/14787210.2025.2484284. Epub 2025 Mar 26.
INTRODUCTION: Traditional microbiological diagnostics face challenges in pathogen identification speed and antimicrobial resistance (AMR) evaluation. Artificial intelligence (AI) offers transformative solutions, necessitating a comprehensive review of its applications, advancements, and integration challenges in clinical microbiology. AREAS COVERED: This review examines AI-driven methodologies, including machine learning (ML), deep learning (DL), and convolutional neural networks (CNNs), for enhancing pathogen detection, AMR prediction, and diagnostic imaging. Applications in virology (e.g. COVID-19 RT-PCR optimization), parasitology (e.g. malaria detection), and bacteriology (e.g. automated colony counting) are analyzed. A literature search was conducted using PubMed, Scopus, and Web of Science (2018-2024), prioritizing peer-reviewed studies on AI's diagnostic accuracy, workflow efficiency, and clinical validation. EXPERT OPINION: AI significantly improves diagnostic precision and operational efficiency but requires robust validation to address data heterogeneity, model interpretability, and ethical concerns. Future success hinges on interdisciplinary collaboration to develop standardized, equitable AI tools tailored for global healthcare settings. Advancing explainable AI and federated learning frameworks will be critical for bridging current implementation gaps and maximizing AI's potential in combating infectious diseases.
100How to build the virtual cell with artificial intelligence: Priorities and opportunities.PubMed
Charlotte Bunne, Yusuf Roohani, Yanay Rosen, et al.
Cell. 2024 Dec 12;187(25):7045-7063. doi: 10.1016/j.cell.2024.11.015.
Cells are essential to understanding health and disease, yet traditional models fall short of modeling and simulating their function and behavior. Advances in AI and omics offer groundbreaking opportunities to create an AI virtual cell (AIVC), a multi-scale, multi-modal large-neural-network-based model that can represent and simulate the behavior of molecules, cells, and tissues across diverse states. This Perspective provides a vision on their design and how collaborative efforts to build AIVCs will transform biological research by allowing high-fidelity simulations, accelerating discoveries, and guiding experimental studies, offering new opportunities for understanding cellular functions and fostering interdisciplinary collaborations in open science.
101A molecular signaling map and its application.PubMed
Jian Li, Ulrich R Mansmann
Cell Signal. 2014 Dec;26(12):2834-42. doi: 10.1016/j.cellsig.2014.08.022. Epub 2014 Sep 2.
Cancer research over the past decades has revealed a number of molecular, biochemical, and cellular events that reflect progressive transformation of normal human cells into their malignant derivatives. These findings help to better understand the complexity of human tumorigenesis. In our study, molecular information is organized to chart a comprehensive map of the signaling network for human cancer. It includes transcriptional and translational regulation and diverse feedback-control loops. It is demonstrated that applying this signaling network map allows predicting the effect of targeted therapy before it can be applied into practice to reduce clinical trial risks. Hence, the proposed map with prognosticating potential effect might become part of drug discovery programs for targeted therapy. Applied in individual patient care it helps to reduce the current reliance of cancer treatment on chemotherapies with low therapeutic indices. This study also demonstrates that continuing elucidation of tumorigenesis will not only need heterotypic organ culture systems in vitro and increasingly refined animal models in vivo, but also computationally calculable virtual cell models in silico.
102Artificial intelligence-based multi-omics analysis fuels cancer precision medicine.PubMed
Xiujing He, Xiaowei Liu, Fengli Zuo, et al.
Semin Cancer Biol. 2023 Jan;88:187-200. doi: 10.1016/j.semcancer.2022.12.009. Epub 2022 Dec 31.
With biotechnological advancements, innovative omics technologies are constantly emerging that have enabled researchers to access multi-layer information from the genome, epigenome, transcriptome, proteome, metabolome, and more. A wealth of omics technologies, including bulk and single-cell omics approaches, have empowered to characterize different molecular layers at unprecedented scale and resolution, providing a holistic view of tumor behavior. Multi-omics analysis allows systematic interrogation of various molecular information at each biological layer while posing tricky challenges regarding how to extract valuable insights from the exponentially increasing amount of multi-omics data. Therefore, efficient algorithms are needed to reduce the dimensionality of the data while simultaneously dissecting the mysteries behind the complex biological processes of cancer. Artificial intelligence has demonstrated the ability to analyze complementary multi-modal data streams within the oncology realm. The coincident development of multi-omics technologies and artificial intelligence algorithms has fuelled the development of cancer precision medicine. Here, we present state-of-the-art omics technologies and outline a roadmap of multi-omics integration analysis using an artificial intelligence strategy. The advances made using artificial intelligence-based multi-omics approaches are described, especially concerning early cancer screening, diagnosis, response assessment, and prognosis prediction. Finally, we discuss the challenges faced in multi-omics analysis, along with tentative future trends in this field. With the increasing application of artificial intelligence in multi-omics analysis, we anticipate a shifting paradigm in precision medicine becoming driven by artificial intelligence-based multi-omics technologies.
103Artificial intelligence for proteomics and biomarker discovery.PubMed
Matthias Mann, Chanchal Kumar, Wen-Feng Zeng, et al.
Cell Syst. 2021 Aug 18;12(8):759-770. doi: 10.1016/j.cels.2021.06.006.
There is an avalanche of biomedical data generation and a parallel expansion in computational capabilities to analyze and make sense of these data. Starting with genome sequencing and widely employed deep sequencing technologies, these trends have now taken hold in all omics disciplines and increasingly call for multi-omics integration as well as data interpretation by artificial intelligence technologies. Here, we focus on mass spectrometry (MS)-based proteomics and describe how machine learning and, in particular, deep learning now predicts experimental peptide measurements from amino acid sequences alone. This will dramatically improve the quality and reliability of analytical workflows because experimental results should agree with predictions in a multi-dimensional data landscape. Machine learning has also become central to biomarker discovery from proteomics data, which now starts to outperform existing best-in-class assays. Finally, we discuss model transparency and explainability and data privacy that are required to deploy MS-based biomarkers in clinical settings.
104Using machine learning approaches for multi-omics data analysis: A review.PubMed
Parminder S Reel, Smarti Reel, Ewan Pearson, et al.
Biotechnol Adv. 2021 Jul-Aug;49:107739. doi: 10.1016/j.biotechadv.2021.107739. Epub 2021 Mar 29.
With the development of modern high-throughput omic measurement platforms, it has become essential for biomedical studies to undertake an integrative (combined) approach to fully utilise these data to gain insights into biological systems. Data from various omics sources such as genetics, proteomics, and metabolomics can be integrated to unravel the intricate working of systems biology using machine learning-based predictive algorithms. Machine learning methods offer novel techniques to integrate and analyse the various omics data enabling the discovery of new biomarkers. These biomarkers have the potential to help in accurate disease prediction, patient stratification and delivery of precision medicine. This review paper explores different integrative machine learning methods which have been used to provide an in-depth understanding of biological systems during normal physiological functioning and in the presence of a disease. It provides insight and recommendations for interdisciplinary professionals who envisage employing machine learning skills in multi-omics studies.
105Multiscale modeling meets machine learning: What can we learn?PubMed
Grace C Y Peng, Mark Alber, Adrian Buganza Tepole, et al.
Arch Comput Methods Eng. 2021 May;28(3):1017-1037. doi: 10.1007/s11831-020-09405-5. Epub 2020 Feb 17.
Machine learning is increasingly recognized as a promising technology in the biological, biomedical, and behavioral sciences. There can be no argument that this technique is incredibly successful in image recognition with immediate applications in diagnostics including electrophysiology, radiology, or pathology, where we have access to massive amounts of annotated data. However, machine learning often performs poorly in prognosis, especially when dealing with sparse data. This is a field where classical physics-based simulation seems to remain irreplaceable. In this review, we identify areas in the biomedical sciences where machine learning and multiscale modeling can mutually benefit from one another: Machine learning can integrate physics-based knowledge in the form of governing equations, boundary conditions, or constraints to manage ill-posted problems and robustly handle sparse and noisy data; multiscale modeling can integrate machine learning to create surrogate models, identify system dynamics and parameters, analyze sensitivities, and quantify uncertainty to bridge the scales and understand the emergence of function. With a view towards applications in the life sciences, we discuss the state of the art of combining machine learning and multiscale modeling, identify applications and opportunities, raise open questions, and address potential challenges and limitations. We anticipate that it will stimulate discussion within the community of computational mechanics and reach out to other disciplines including mathematics, statistics, computer science, artificial intelligence, biomedicine, systems biology, and precision medicine to join forces towards creating robust and efficient models for biological systems.
107Multiscale Modeling in Computational Biomechanics: A New Era with Virtual Human Twins and Contemporary Artificial Intelligence.PubMed
Tien-Tuan Dao
Bioengineering (Basel). 2025 Mar 20;12(3):320. doi: 10.3390/bioengineering12030320.
Over the last several decades, computational biomechanics has been intensively investigated as part of the study of human body systems (musculoskeletal, cardiovascular, digestive, etc [...].
108DeepProg: an ensemble of deep-learning and machine-learning models for prognosis prediction using multi-omics data.PubMed
Olivier B Poirion, Zheng Jing, Kumardeep Chaudhary, et al.
Genome Med. 2021 Jul 14;13(1):112. doi: 10.1186/s13073-021-00930-x.
Multi-omics data are good resources for prognosis and survival prediction; however, these are difficult to integrate computationally. We introduce DeepProg, a novel ensemble framework of deep-learning and machine-learning approaches that robustly predicts patient survival subtypes using multi-omics data. It identifies two optimal survival subtypes in most cancers and yields significantly better risk-stratification than other multi-omics integration methods. DeepProg is highly predictive, exemplified by two liver cancer (C-index 0.73-0.80) and five breast cancer datasets (C-index 0.68-0.73). Pan-cancer analysis associates common genomic signatures in poor survival subtypes with extracellular matrix modeling, immune deregulation, and mitosis processes. DeepProg is freely available at https://github.com/lanagarmire/DeepProg.
109A roadmap for multi-omics data integration using deep learning.PubMed
Mingon Kang, Euiseong Ko, Tesfaye B Mersha
Brief Bioinform. 2022 Jan 17;23(1). doi: 10.1093/bib/bbab454.
High-throughput next-generation sequencing now makes it possible to generate a vast amount of multi-omics data for various applications. These data have revolutionized biomedical research by providing a more comprehensive understanding of the biological systems and molecular mechanisms of disease development. Recently, deep learning (DL) algorithms have become one of the most promising methods in multi-omics data analysis, due to their predictive performance and capability of capturing nonlinear and hierarchical features. While integrating and translating multi-omics data into useful functional insights remain the biggest bottleneck, there is a clear trend towards incorporating multi-omics analysis in biomedical research to help explain the complex relationships between molecular layers. Multi-omics data have a role to improve prevention, early detection and prediction; monitor progression; interpret patterns and endotyping; and design personalized treatments. In this review, we outline a roadmap of multi-omics integration using DL and offer a practical perspective into the advantages, challenges and barriers to the implementation of DL in multi-omics data.
110The Impact of Artificial Intelligence on Traditional Chinese Medicine.PubMed
Yulin Wang, Xiuming Shi, Li Li, et al.
Am J Chin Med. 2021;49(6):1297-1314. doi: 10.1142/S0192415X21500622. Epub 2021 Jul 10.
Traditional Chinese Medicine (TCM) is a well-established medical system with a long history. Currently, artificial intelligence (AI) is rapidly expanding in many fields including TCM. AI will significantly improve the reliability and accuracy of diagnostics, thus increasing the use of effective therapeutic methods for patients. This systematic review provides an updated overview on the major breakthroughs in the field of AI-assisted TCM four diagnostic methods, syndrome differentiation, and treatment. AI-assisted TCM diagnosis is mainly based on digital data collected by modern electronic instruments, which makes TCM diagnosis more quantitative, objective, and standardized. As a result, the diagnosis decisions made by different TCM doctors exhibit more consistency, accuracy, and reliability. Meanwhile, the therapeutic efficacy of TCM can be evaluated objectively. Therefore, AI is promoting TCM from experience to evidence-based medicine, a genuine scientific revolution. Furthermore, huge and non-uniform knowledge on formula-syndrome relationships and the combination rules of herbal TCM formulae could be better standardized with the help of AI analysis, which is necessary for the clinical efficacy evaluation and further optimization on the standardized TCM formulae. AI bridges the gap between TCM and modern science and technology. AI may bring clinical TCM diagnostics closer to western medicine. With the help of AI, more scientific evidence about TCM will be discovered. It can be expected that more unified guidelines for specific TCM syndromes will be issued with the development of AI-assisted TCM therapies in the future.
111Modern bioinformatics meets traditional Chinese medicine.PubMed
Peiqin Gu, Huajun Chen
Brief Bioinform. 2014 Nov;15(6):984-1003. doi: 10.1093/bib/bbt063. Epub 2013 Sep 24.
Traditional Chinese medicine (TCM) is gaining increasing attention with the emergence of integrative medicine and personalized medicine, characterized by pattern differentiation on individual variance and treatments based on natural herbal synergism. Investigating the effectiveness and safety of the potential mechanisms of TCM and the combination principles of drug therapies will bridge the cultural gap with Western medicine and improve the development of integrative medicine. Dealing with rapidly growing amounts of biomedical data and their heterogeneous nature are two important tasks among modern biomedical communities. Bioinformatics, as an emerging interdisciplinary field of computer science and biology, has become a useful tool for easing the data deluge pressure by automating the computation processes with informatics methods. Using these methods to retrieve, store and analyze the biomedical data can effectively reveal the associated knowledge hidden in the data, and thus promote the discovery of integrated information. Recently, these techniques of bioinformatics have been used for facilitating the interactional effects of both Western medicine and TCM. The analysis of TCM data using computational technologies provides biological evidence for the basic understanding of TCM mechanisms, safety and efficacy of TCM treatments. At the same time, the carrier and targets associated with TCM remedies can inspire the rethinking of modern drug development. This review summarizes the significant achievements of applying bioinformatics techniques to many aspects of the research in TCM, such as analysis of TCM-related '-omics' data and techniques for analyzing biological processes and pharmaceutical mechanisms of TCM, which have shown certain potential of bringing new thoughts to both sides.
112An AI-driven strategy for active compounds discovery and non-destructive quality control in traditional Chinese medicine: A case of Xuefu Zhuyu Oral Liquid.PubMed
Lele Gao, Liang Zhong, Tingting Feng, et al.
Talanta. 2025 May 15;287:127627. doi: 10.1016/j.talanta.2025.127627. Epub 2025 Jan 23.
The modernization and globalization of traditional Chinese medicine (TCM) face challenges such as unclear active compounds and inadequate quality control. Taking Xuefu Zhuyu Oral Liquid (XZOL) as an example, this study proposed an artificial intelligence (AI) -driven strategy for active compounds discovery and non-destructive quality control. Firstly, the multi-wavelength fusion high-performance liquid chromatography (HPLC) fingerprints were constructed to comprehensively characterize the chemical composition of XZOL. Secondly, the pro-angiogenesis effects of XZOL were evaluated in a PTK787-induced intersegmental vessels (ISVs) injury zebrafish model. Then, spectrum-effect relationship models, incorporating gray relational analysis (GRA), partial least squares regression (PLSR), backpropagation artificial neural networks (BP-ANN), and convolutional neural networks (CNN), discovered seven pro-angiogenesis active compounds (Hydroxysafflor Yellow A, Paeoniflorin, Ferulic Acid, Narirutin, Naringin, Hesperidin, and Neohesperidin). Furthermore, the efficacy of these compounds was further validated through network pharmacology, molecular docking, and zebrafish. Finally, a rapid and non-destructive quality control system based on near infrared spectroscopy (NIRS) was established. This system effectively distinguished expired and normal samples by combining Hotelling T and Distance to Model X (DModX) statistics of multivariate statistical process control (MSPC), and accurately predicted the content of above active compounds by CNN model integration with bidirectional long short-term memory (Bi-LSTM) and multi-head self-attention (MHSA) networks. This study underscores the potential of AI-driven strategy to enhance TCM standardization and global recognition by providing an active compounds-based holistic quality control strategy of TCM.
113Network pharmacology prediction and molecular docking-based strategy to explore the potential mechanism of Duhuo-Jisheng pair against osteoarthritis.PubMed
Haiyang Kou, Jianbing Ma
J Pharm Pharmacol. 2024 Jul 5;76(7):743-752. doi: 10.1093/jpp/rgad117.
OBJECTIVE: The Duhuo-Jisheng pair is the ruling herb in Duhuo Jisheng decoction, which is a classic formula first recorded in the preparedness and urgency of the thousand jewels. METHODS: We obtained the primary constituents of Duhuo-Jisheng and their associated protein targets from the TCMSP database. We constructed a composite target network using Cytoscape 3.9.1. To identify potential targets for the treatment of osteoarthritis (OA), we retrieved disease targets from OMIM and GeneCards databases and compared them with the composite targets. We imported the overlapping targets into the STRING database to construct a protein-protein interaction (PPI) network. We also conducted Gene ontology (GO) and KEGG enrichment analyses on the targets. RESULTS: The component target network consisted of numerous nodes and edges. Notably, quercetin, ammidin, and β-sitosterol were identified as the compounds with high degrees. The PPI network identified tumour necrosis factor (TNF), TP53, and NOS2 as proteins with high degrees. The results of GO and KEGG analyses revealed that the signalling pathways used by DHQJD to treat OA included the NF-κB, PI3K-AKT, and TNF pathways. CONCLUSION: Our study provides insights into the effective components and potential molecular mechanisms of Duhuo-Jisheng in treating OA, thus serving as a reference for further basic research in this field.
114Exploring therapeutic targets and molecular mechanisms for treating diabetes mellitus-associated heart failure with Qishen Yiqi dropping pills: A network pharmacology and bioinformatics approach.PubMed
Yirong Ma, Junyu Lai, Zhengtao Chen, et al.
Medicine (Baltimore). 2024 Aug 2;103(31):e39104. doi: 10.1097/MD.0000000000039104.
Diabetes mellitus (DM) and heart failure frequently coexist, presenting significant public health challenges. QiShenYiQi Dropping Pills (QSDP) are widely employed in the treatment of diabetes mellitus concomitant with heart failure (DM-HF). Nevertheless, the precise mechanisms underlying their efficacy have yet to be elucidated. Active ingredients and likely targets of QSDP were retrieved from the TCMSP and UniProt databases. Genes associated with DM-HF were pinpointed through searches in the GeneCards, OMIM, DisGeNET, and TTD databases. Differential genes connected to DM-HF were sourced from the GEO database. Enrichment analyses via gene ontology and Kyoto Encyclopedia of Genes and Genomes pathways, as well as immune infiltration assessments, were conducted using R software. Further analysis involved employing molecular docking strategies to explore the interactions between the identified targets and active substances in QSDP that are pertinent to DM-HF treatment. This investigation effectively discerned 108 active compounds and 257 targets relevant to QSDP. A protein-protein interaction network was constructed, highlighting 6 central targets for DM-HF treatment via QSDP. Gene ontology enrichment analysis predominantly linked these targets with responses to hypoxia, metabolism of reactive oxygen species, and cytokine receptor interactions. Analysis of Kyoto Encyclopedia of Genes and Genomes pathways demonstrated that these targets mainly participate in pathways linked to diabetic complications, such as AGE-RAGE signaling, dyslipidemia, arteriosclerosis, the HIF-1 signaling pathway, and the tumor necrosis factor signaling pathway. Further, immune infiltration analysis implied that QSDP's mechanism in treating DM-HF might involve immune-mediated inflammation and crucial signaling pathways. Additionally, molecular docking studies showed that the active substances in QSDP have strong binding affinities with these identified targets. This research presents a new model for addressing DM-HF through the use of QSDP, providing novel insights into incorporating traditional Chinese medicine (TCM) principles in the clinical treatment of DM-HF. The implications of these findings are substantial for both clinical application and further scientific inquiry.
115Uncovering the mechanism of Ge-Gen-Qin-Lian decoction for treating ulcerative colitis based on network pharmacology and molecular docking verification.PubMed
Lin Xu, Jiaqi Zhang, Yifan Wang, et al.
Biosci Rep. 2021 Feb 26;41(2). doi: 10.1042/BSR20203565.
BACKGROUND: Ge-Gen-Qin-Lian Decoction (GGQLD), a traditional Chinese medicine (TCM) formula, has been widely used for ulcerative colitis (UC) in China, but the pharmacological mechanisms remain unclear. This research was designed to clarify the underlying pharmacological mechanism of GGQLD against UC. METHOD: In this research, a GGQLD-compound-target-UC network was constructed based on public databases to clarify the relationship between active compounds in GGQLD and potential targets. Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analyses were performed to investigate biological functions associated with potential targets. A protein-protein interaction network was constructed to screen and evaluate hub genes and key active ingredients. Molecular docking was used to verify the activities of binding between hub targets and ingredients. RESULTS: Finally, 83 potential therapeutic targets and 118 corresponding active ingredients were obtained by network pharmacology. Quercetin, kaempferol, wogonin, baicalein, and naringenin were identified as potential candidate ingredients. GO and KEGG enrichment analyses revealed that GGQLD had anti-inflammatory, antioxidative, and immunomodulatory effects. The effect of GGQLD on UC might be achieved by regulating the balance of cytokines (e.g., IL-6, TNF, IL-1β, CXCL8, CCL2) in the immune system and inflammation-related pathways, such as the IL-17 pathway and the Th17 cell differentiation pathway. In addition, molecular docking results demonstrated that the main active ingredient, quercetin, exhibited good affinity to hub targets. CONCLUSION: This research fully reflects the multicomponent and multitarget characteristics of GGQLD in the treatment of UC. Furthermore, the present study provided new insight into the mechanisms of GGQLD against UC.
116Traditional Chinese Medicine for Anti-Arrhythmias: Mechanisms via Potassium Channels.PubMed
Chenhao Gao, Ting Xu, Lingyu Linda Ye, et al.
Basic Clin Pharmacol Toxicol. 2025 Jul;137(1):e70059. doi: 10.1111/bcpt.70059.
Cardiac arrhythmia is a common life-threatening cardiovascular disorder. Potassium channels play a crucial role in cardiac electrophysiology, and their dysfunction is closely associated with the occurrence and development of arrhythmia. Traditional Chinese medicine (TCM) has a long-standing history of treating various diseases, including arrhythmia and offers a rich source of compounds for anti-arrhythmic drugs. To provide in-depth new insights for the development of novel anti-arrhythmic therapies, this review delves into the effects of TCM on potassium channels through multiple ways, including the direct interaction between the active ingredients contained and potassium channels, regulation of the expression of related genes and proteins, intervention in intracellular signalling pathways and exertion of antioxidant and anti-inflammatory effects. With continued efforts in standardization, research and integration with Western medicine, TCM is playing an increasingly important role in the treatment of arrhythmia and provides novel options for patients worldwide. Phenomics and pharmacophenomics as well as the artificial intelligence-based GPCR ligand screening system may provide new paradigms and platforms for discovering and developing more anti-arrhythmic drugs.
117Advances in the Application of Traditional Chinese Medicine Using Artificial Intelligence: A Review.PubMed
Sheng Zhang, Wei Wang, Xitian Pi, et al.
Am J Chin Med. 2023;51(5):1067-1083. doi: 10.1142/S0192415X23500490. Epub 2023 Jul 7.
Traditional Chinese medicine (TCM), as one of the crystallizations of Chinese wisdom, emphasizes the balance of Yin and Yang to keep the body healthy. Under the theoretical guidance of a holistic view, the diagnostic process in TCM has characteristics of subjectivity, fuzziness, and complexity. Therefore, realizing standardization and achieving objective quantitative analysis are the bottlenecks of the development of TCM. The emergence of artificial intelligence (AI) technology has brought unprecedented challenges and opportunities to traditional medicine, which is expected to provide objective measurements and improve the clinical efficacy. However, the combination of TCM and AI is still in its infancy and currently faces many challenges. Therefore, this review provides a comprehensive discussion of the existing advances, problems, and prospects of the applications of AI technologies in TCM with the hope of promoting a better understanding of the TCM modernization and intellectualization.
118Development and Application of Traditional Chinese Medicine Using AI Machine Learning and Deep Learning Strategies.PubMed
Danping Pan, Yilei Guo, Yongfu Fan, et al.
Am J Chin Med. 2024;52(3):605-623. doi: 10.1142/S0192415X24500265. Epub 2024 May 8.
Traditional Chinese medicine (TCM) has been used for thousands of years and has been proven to be effective at treating many complicated illnesses with minimal side effects. The application and advancement of TCM are, however, constrained by the absence of objective measuring standards due to its relatively abstract diagnostic methods and syndrome differentiation theories. Ongoing developments in machine learning (ML) and deep learning (DL), specifically in computer vision (CV) and natural language processing (NLP), offer novel opportunities to modernize TCM by exploring the profound connotations of its theory. This review begins with an overview of the ML and DL methods employed in TCM; this is followed by practical instances of these applications. Furthermore, extensive discussions emphasize the mature integration of ML and DL in TCM, such as tongue diagnosis, pulse diagnosis, and syndrome differentiation treatment, highlighting their early successful application in the TCM field. Finally, this study validates the accomplishments and addresses the problems and challenges posed by the application and development of TCM powered by ML and DL. As ML and DL techniques continue to evolve, modern technology will spark new advances in TCM.
119Decoding aging clocks: New insights from metabolomics.PubMed
Honghao Huang, Yifan Chen, Wei Xu, et al.
Cell Metab. 2025 Jan 7;37(1):34-58. doi: 10.1016/j.cmet.2024.11.007. Epub 2024 Dec 9.
Chronological age is a crucial risk factor for diseases and disabilities among older adults. However, individuals of the same chronological age often exhibit divergent biological aging states, resulting in distinct individual risk profiles. Chronological age estimators based on omics data and machine learning techniques, known as aging clocks, provide a valuable framework for interpreting molecular-level biological aging. Metabolomics is an intriguing and rapidly growing field of study, involving the comprehensive profiling of small molecules within the body and providing the ultimate genome-environment interaction readout. Consequently, leveraging metabolomics to characterize biological aging holds immense potential. The aim of this review was to provide an overview of metabolomics approaches, highlighting the establishment and interpretation of metabolomic aging clocks while emphasizing their strengths, limitations, and applications, and to discuss their underlying biological significance, which has the potential to drive innovation in longevity research and development.
120Biomarkers of aging in frailty and age-associated disorders: State of the art and future perspective.PubMed
Stefano Salvioli, Maria Sofia Basile, Leonardo Bencivenga, et al.
Ageing Res Rev. 2023 Nov;91:102044. doi: 10.1016/j.arr.2023.102044. Epub 2023 Aug 28.
According to the Geroscience concept that organismal aging and age-associated diseases share the same basic molecular mechanisms, the identification of biomarkers of age that can efficiently classify people as biologically older (or younger) than their chronological (i.e. calendar) age is becoming of paramount importance. These people will be in fact at higher (or lower) risk for many different age-associated diseases, including cardiovascular diseases, neurodegeneration, cancer, etc. In turn, patients suffering from these diseases are biologically older than healthy age-matched individuals. Many biomarkers that correlate with age have been described so far. The aim of the present review is to discuss the usefulness of some of these biomarkers (especially soluble, circulating ones) in order to identify frail patients, possibly before the appearance of clinical symptoms, as well as patients at risk for age-associated diseases. An overview of selected biomarkers will be discussed in this regard, in particular we will focus on biomarkers related to metabolic stress response, inflammation, and cell death (in particular in neurodegeneration), all phenomena connected to inflammaging (chronic, low-grade, age-associated inflammation). In the second part of the review, next-generation markers such as extracellular vesicles and their cargos, epigenetic markers and gut microbiota composition, will be discussed. Since recent progresses in omics techniques have allowed an exponential increase in the production of laboratory data also in the field of biomarkers of age, making it difficult to extract biological meaning from the huge mass of available data, Artificial Intelligence (AI) approaches will be discussed as an increasingly important strategy for extracting knowledge from raw data and providing practitioners with actionable information to treat patients.
121Decoding disease-specific ageing mechanisms through pathway-level epigenetic clock: insights from multi-cohort validation.PubMed
Pan Li, Jijun Zhu, Shenghan Wang, et al.
EBioMedicine. 2025 Aug;118:105829. doi: 10.1016/j.ebiom.2025.105829. Epub 2025 Jul 5.
BACKGROUND: Ageing is a multifactorial process closely associated with increased risk of chronic diseases. While epigenetic clocks have advanced ageing research, most rely on isolated CpG sites, limiting biological interpretability. We developed PathwayAge, a biologically informed model that captures coordinated methylation changes at the pathway level, providing interpretable insights into ageing biology and disease mechanisms. METHODS: We conducted a cross-sectional study using genome-wide DNA methylation data from 10,615 individuals across 19 cohorts and 3413 Han Chinese participants, along with transcriptomic data from 3384 samples. A two-stage machine learning model aggregated CpG sites into GO or KEGG pathway-level features to predict chronological age. Model accuracy was assessed using mean absolute error (MAE) and Pearson correlation (Rho). Age acceleration residuals (AgeAcc) were computed and tested for associations with nine diseases using non-parametric statistics. FINDINGS: PathwayAge achieved high predictive accuracy (Rho = 0.977, MAE = 2.350) in cross-validation and across 15 independent blood-based validation cohorts (Rho = 0.677-0.979, MAE = 2.113-6.837), including a Chinese population (Rho = 0.972, MAE = 2.302). Compared to established clocks, PathwayAge showed improved performance in both age estimation and disease association analyses. Significant AgeAcc differences were observed across nine diseases, with disease-specific pathways confirmed by permutation tests (P < 0.02). Top pathways implicated in ageing included autophagy, cell adhesion, synaptic signalling, and metabolic regulation. GO-based clustering revealed consistent ageing signatures across disease categories, including neuropsychiatric, immune, metabolic, and cancer-related conditions. Cross-omics validation using transcriptomic data further supported the model's biological relevance (Rho = 0.70, MAE = 7.21). INTERPRETATION: PathwayAge represents an interpretable, biologically grounded framework for estimating epigenetic age. By integrating pathway-level methylation signals, it uncovers mechanistic links between ageing and disease, with potential applications in biomarker development and precision ageing medicine. FUNDING: This research was supported by the Greater Bay Area Institute of Precision Medicine (Grant No. I0007), the National Social Science Foundation of China (Grant No. 32370639), and was further supported by the Shanghai Key Laboratory of Psychotic Disorders Open Grant (Grant No: 21-K01). ES received funding from the Hector II Foundation and the German Federal Ministry of Education and Research (BEST project, Grant No: 01EK2101B), and was endorsed by the German Center for Mental Health (DZPG). ES received speaker fees from bfd Buchholz-Fachinformationsdienst GmbH and editorial fees from the Lundbeck Foundation. SW received funding from the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant no. XDB38020400), CAS Young Team Program for Stable Support of Basic Research (YSBR-077), CAS Interdisciplinary Innovation Team, Shanghai Municipal Science and Technology Major Project (Grant No. 2017SHZDZX01 to SW), the National Natural Science Foundation of China (32325013 and 92249302), the National Key Research and Development Project (2018YFC0910403), Shanghai Science and Technology Commission Excellent Academic Leaders Program (22XD1424700).
122Acceleration of the DNA methylation clock among lynch syndrome-associated mutation carriers.PubMed
Marta Cuadros, Carlos Cano, Sonia Garcia-Rodriguez, et al.
BMC Med Genomics. 2022 Mar 4;15(1):45. doi: 10.1186/s12920-022-01183-2.
BACKGROUND: DNA methylation (DNAm) age metrics have been widely accepted as an epigenetic biomarker for biological aging and disease. The purpose of this study is to assess whether or not individuals carrying Lynch Syndrome-associated mutations are affected in their rate of biological aging, as measured by the epigenetic clock. METHODS: Genome-wide bisulfite DNA sequencing data were generated using DNA from CD4 + T-cells obtained from peripheral blood using 27 patient samples from Lynch syndrome families. Horvath's DNAm age model based on penalized linear regression was applied to estimate DNAm age from patient samples with distinct clinical and genetic characteristics to investigate cancer mutation-related aging effects. RESULTS: Both Lynch mutation carriers and controls exhibited high variability in their estimated DNAm age, but regression analysis showed steeper slope for the Lynch mutation carriers. Remarkably, six Lynch Syndrome-associated mutation carriers showed a strong correlation to the control group, and two sisters carrying Lynch Syndrome-associated mutations, with no significant difference in lifestyle and similar chronological age, were assigned very different DNAm age. CONCLUSIONS: Future studies will be required to explore, in larger patient populations, whether specific epigenetic age acceleration is predictive of time-to-cancer development, treatment response, and survival. Epigenetic clock DNAm metrics may be affected by the presence of cancer mutations in the germline, and thus show promise of potential clinical utility for stratified surveillance strategies based on the relative risk for imminent emergence of tumor lesions in otherwise healthy Lynch Syndrome-associated mutation carriers.
123Artificial Intelligence-Driven Biological Age Prediction Model Using Comprehensive Health Checkup Data: Development and Validation Study.PubMed
Chang-Uk Jeong, Jacob S Leiby, Dokyoon Kim, et al.
JMIR Aging. 2025 Apr 11;8:e64473. doi: 10.2196/64473.
BACKGROUND: The global increase in life expectancy has not shown a similar rise in healthy life expectancy. Accurate assessment of biological aging is crucial for mitigating diseases and socioeconomic burdens associated with aging. Current biological age prediction models are limited by their reliance on conventional statistical methods and constrained clinical information. OBJECTIVE: This study aimed to develop and validate an aging clock model using artificial intelligence, based on comprehensive health check-up data, to predict biological age and assess its clinical relevance. METHODS: We used data from Koreans who underwent health checkups at the Seoul National University Hospital Gangnam Center as well as from the Korean Genome and Epidemiology Study. Our model incorporated 27 clinical factors and employed machine learning algorithms, including linear regression, least absolute shrinkage and selection operator, ridge regression, elastic net, random forest, support vector machine, gradient boosting, and K-nearest neighbors. Model performance was evaluated using adjusted R2 and the mean squared error (MSE) values. Shapley Additive exPlanation (SHAP) analysis was conducted to interpret the model's predictions. RESULTS: The Gradient Boosting model achieved the best performance with a mean (SE) MSE of 4.219 (0.14) and a mean (SE) R2 of 0.967 (0.001). SHAP analysis identified significant predictors of biological age, including kidney function markers, gender, glycated hemoglobin level, liver function markers, and anthropometric measurements. After adjusting for the chronological age, the predicted biological age showed strong associations with multiple clinical factors, such as metabolic status, body compositions, fatty liver, smoking status, and pulmonary function. CONCLUSIONS: Our aging clock model demonstrates a high predictive accuracy and clinical relevance, offering a valuable tool for personalized health monitoring and intervention. The model's applicability in routine health checkups could enhance health management and promote regular health evaluations.
124Machine Learning Approximations to Predict Epigenetic Age Acceleration in Stroke Patients.PubMed
Isabel Fernández-Pérez, Joan Jiménez-Balado, Uxue Lazcano, et al.
Int J Mol Sci. 2023 Feb 1;24(3):2759. doi: 10.3390/ijms24032759.
Age acceleration (Age-A) is a useful tool that is able to predict a broad range of health outcomes. It is necessary to determine DNA methylation levels to estimate it, and it is known that Age-A is influenced by environmental, lifestyle, and vascular risk factors (VRF). The aim of this study is to estimate the contribution of these easily measurable factors to Age-A in patients with cerebrovascular disease (CVD), using different machine learning (ML) approximations, and try to find a more accessible model able to predict Age-A. We studied a CVD cohort of 952 patients with information about VRF, lifestyle habits, and target organ damage. We estimated Age-A using Hannum's epigenetic clock, and trained six different models to predict Age-A: a conventional linear regression model, four ML models (elastic net regression (EN), K-Nearest neighbors, random forest, and support vector machine models), and one deep learning approximation (multilayer perceptron (MLP) model). The best-performing models were EN and MLP; although, the predictive capability was modest (R 0.358 and 0.378, respectively). In conclusion, our results support the influence of these factors on Age-A; although, they were not enough to explain most of its variability.
125Longitudinal serum proteome mapping reveals biomarkers for healthy ageing and related cardiometabolic diseases.PubMed
Jun Tang, Liang Yue, Ying Xu, et al.
Nat Metab. 2025 Jan;7(1):166-181. doi: 10.1038/s42255-024-01185-7. Epub 2025 Jan 13.
The blood proteome contains biomarkers of ageing and age-associated diseases, but such markers are rarely validated longitudinally. Here we map the longitudinal proteome in 7,565 serum samples from a cohort of 3,796 middle-aged and elderly adults across three time points over a 9-year follow-up period. We pinpoint 86 ageing-related proteins that exhibit signatures associated with 32 clinical traits and the incidence of 14 major ageing-related chronic diseases. Leveraging a machine-learning model, we pick 22 of these proteins to generate a proteomic healthy ageing score (PHAS), capable of predicting the incidence of cardiometabolic diseases. We further identify the gut microbiota as a modifiable factor influencing the PHAS. Our data constitute a valuable resource and offer useful insights into the roles of serum proteins in ageing and age-associated cardiometabolic diseases, providing potential targets for intervention with therapeutics to promote healthy ageing.
126An interpretable machine learning-based cerebrospinal fluid proteomics clock for predicting age reveals novel insights into brain aging.PubMed
Justin Melendez, Yun Ju Sung, Miranda Orr, et al.
Aging Cell. 2024 Sep;23(9):e14230. doi: 10.1111/acel.14230. Epub 2024 Jun 24.
Machine learning can be used to create "biologic clocks" that predict age. However, organs, tissues, and biofluids may age at different rates from the organism as a whole. We sought to understand how cerebrospinal fluid (CSF) changes with age to inform the development of brain aging-related disease mechanisms and identify potential anti-aging therapeutic targets. Several epigenetic clocks exist based on plasma and neuronal tissues; however, plasma may not reflect brain aging specifically and tissue-based clocks require samples that are difficult to obtain from living participants. To address these problems, we developed a machine learning clock that uses CSF proteomics to predict the chronological age of individuals with a 0.79 Pearson correlation and mean estimated error (MAE) of 4.30 years in our validation cohort. Additionally, we analyzed proteins highly weighted by the algorithm to gain insights into changes in CSF and uncover novel insights into brain aging. We also demonstrate a novel method to create a minimal protein clock that uses just 109 protein features from the original clock to achieve a similar accuracy (0.75 correlation, MAE 5.41). Finally, we demonstrate that our clock identifies novel proteins that are highly predictive of age in interactions with other proteins, but do not directly correlate with chronological age themselves. In conclusion, we propose that our CSF protein aging clock can identify novel proteins that influence the rate of aging of the central nervous system (CNS), in a manner that would not be identifiable by examining their individual relationships with age.
127Nonlinear dynamics of multi-omics profiles during human aging.PubMed
Xiaotao Shen, Chuchu Wang, Xin Zhou, et al.
Nat Aging. 2024 Nov;4(11):1619-1634. doi: 10.1038/s43587-024-00692-2. Epub 2024 Aug 14.
Aging is a complex process associated with nearly all diseases. Understanding the molecular changes underlying aging and identifying therapeutic targets for aging-related diseases are crucial for increasing healthspan. Although many studies have explored linear changes during aging, the prevalence of aging-related diseases and mortality risk accelerates after specific time points, indicating the importance of studying nonlinear molecular changes. In this study, we performed comprehensive multi-omics profiling on a longitudinal human cohort of 108 participants, aged between 25 years and 75 years. The participants resided in California, United States, and were tracked for a median period of 1.7 years, with a maximum follow-up duration of 6.8 years. The analysis revealed consistent nonlinear patterns in molecular markers of aging, with substantial dysregulation occurring at two major periods occurring at approximately 44 years and 60 years of chronological age. Distinct molecules and functional pathways associated with these periods were also identified, such as immune regulation and carbohydrate metabolism that shifted during the 60-year transition and cardiovascular disease, lipid and alcohol metabolism changes at the 40-year transition. Overall, this research demonstrates that functions and risks of aging-related diseases change nonlinearly across the human lifespan and provides insights into the molecular and biological pathways involved in these changes.
128Phenome-wide associations of human aging uncover sex-specific dynamics.PubMed
Lee Reicher, Noam Bar, Anastasia Godneva, et al.
Nat Aging. 2024 Nov;4(11):1643-1655. doi: 10.1038/s43587-024-00734-9. Epub 2024 Nov 5.
Aging varies significantly among individuals of the same chronological age, indicating that biological age (BA), estimated from molecular and physiological biomarkers, may better reflect aging. Prior research has often ignored sex-specific differences in aging patterns and mainly focused on aging biomarkers from a single data modality. Here we analyze a deeply phenotyped longitudinal cohort (10K project, Israel) of 10,000 healthy individuals aged 40-70 years that includes clinical, physiological, behavioral, environmental and multiomic parameters. Follow-up visits are scheduled every 2 years for a total of 25 years. We devised machine learning models of chronological age and computed biological aging scores that represented diverse physiological systems, revealing different aging patterns among sexes. Higher BA scores were associated with a higher prevalence of age-related medical conditions, highlighting the clinical relevance of these scores. Our analysis revealed system-specific aging dynamics and the potential of deeply phenotyped cohorts to accelerate improvements in our understanding of chronic diseases. Our findings present a more holistic view of the aging process, and lay the foundation for personalized medical prevention strategies.
129Organ-specific biological clocks: Ageotyping for personalized anti-aging medicine.PubMed
Francesco Prattichizzo, Chiara Frigé, Valeria Pellegrini, et al.
Ageing Res Rev. 2024 Apr;96:102253. doi: 10.1016/j.arr.2024.102253. Epub 2024 Mar 4.
Aging is a complex multidimensional, progressive remodeling process affecting multiple organ systems. While many studies have focused on studying aging across multiple organs, assessment of the contribution of individual organs to overall aging processes is a cutting-edge issue. An organ's biological age might influence the aging of other organs, revealing a multiorgan aging network. Recent data demonstrated a similar yet asynchronous inter-organs and inter-individuals progression of aging, thereby providing a foundation to track sources of declining health in old age. The integration of multiple omics with common clinical parameters through artificial intelligence has allowed the building of organ-specific aging clocks, which can predict the development of specific age-related diseases at high resolution. The peculiar individual aging-trajectory, referred to as ageotype, might provide a novel tool for a personalized anti-aging, preventive medicine. Here, we review data relative to biological aging clocks and omics-based data, suggesting different organ-specific aging rates. Additional research on longitudinal data, including young subjects and analyzing sex-related differences, should be encouraged to apply ageotyping analysis for preventive purposes in clinical practice.
130Spatiotemporal transcriptomic changes of human ovarian aging and the regulatory role of FOXP1.PubMed
Meng Wu, Weicheng Tang, Ying Chen, et al.
Nat Aging. 2024 Apr;4(4):527-545. doi: 10.1038/s43587-024-00607-1. Epub 2024 Apr 9.
Limited understanding exists regarding how aging impacts the cellular and molecular aspects of the human ovary. This study combines single-cell RNA sequencing and spatial transcriptomics to systematically characterize human ovarian aging. Spatiotemporal molecular signatures of the eight types of ovarian cells during aging are observed. An analysis of age-associated changes in gene expression reveals that DNA damage response may be a key biological pathway in oocyte aging. Three granulosa cells subtypes and five theca and stromal cells subtypes, as well as their spatiotemporal transcriptomics changes during aging, are identified. FOXP1 emerges as a regulator of ovarian aging, declining with age and inhibiting CDKN1A transcription. Silencing FOXP1 results in premature ovarian insufficiency in mice. These findings offer a comprehensive understanding of spatiotemporal variability in human ovarian aging, aiding the prioritization of potential diagnostic biomarkers and therapeutic strategies.
131Longevity biotechnology: bridging AI, biomarkers, geroscience and clinical applications for healthy longevity.PubMed
Yu-Xuan Lyu, Qiang Fu, Dominika Wilczok, et al.
Aging (Albany NY). 2024 Oct 16;16(20):12955-12976. doi: 10.18632/aging.206135.
The recent unprecedented progress in ageing research and drug discovery brings together fundamental research and clinical applications to advance the goal of promoting healthy longevity in the human population. We, from the gathering at the Aging Research and Drug Discovery Meeting in 2023, summarised the latest developments in healthspan biotechnology, with a particular emphasis on artificial intelligence (AI), biomarkers and clocks, geroscience, and clinical trials and interventions for healthy longevity. Moreover, we provide an overview of academic research and the biotech industry focused on targeting ageing as the root of age-related diseases to combat multimorbidity and extend healthspan. We propose that the integration of generative AI, cutting-edge biological technology, and longevity medicine is essential for extending the productive and healthy human lifespan.
132Artificial intelligence for aging and longevity research: Recent advances and perspectives.PubMed
Alex Zhavoronkov, Polina Mamoshina, Quentin Vanhaelen, et al.
Ageing Res Rev. 2019 Jan;49:49-66. doi: 10.1016/j.arr.2018.11.003. Epub 2018 Nov 22.
The applications of modern artificial intelligence (AI) algorithms within the field of aging research offer tremendous opportunities. Aging is an almost universal unifying feature possessed by all living organisms, tissues, and cells. Modern deep learning techniques used to develop age predictors offer new possibilities for formerly incompatible dynamic and static data types. AI biomarkers of aging enable a holistic view of biological processes and allow for novel methods for building causal models-extracting the most important features and identifying biological targets and mechanisms. Recent developments in generative adversarial networks (GANs) and reinforcement learning (RL) permit the generation of diverse synthetic molecular and patient data, identification of novel biological targets, and generation of novel molecular compounds with desired properties and geroprotectors. These novel techniques can be combined into a unified, seamless end-to-end biomarker development, target identification, drug discovery and real world evidence pipeline that may help accelerate and improve pharmaceutical research and development practices. Modern AI is therefore expected to contribute to the credibility and prominence of longevity biotechnology in the healthcare and pharmaceutical industry, and to the convergence of countless areas of research.
133Discovery of senolytics using machine learning.PubMed
Vanessa Smer-Barreto, Andrea Quintanilla, Richard J R Elliott, et al.
Nat Commun. 2023 Jun 10;14(1):3445. doi: 10.1038/s41467-023-39120-1.
Cellular senescence is a stress response involved in ageing and diverse disease processes including cancer, type-2 diabetes, osteoarthritis and viral infection. Despite growing interest in targeted elimination of senescent cells, only few senolytics are known due to the lack of well-characterised molecular targets. Here, we report the discovery of three senolytics using cost-effective machine learning algorithms trained solely on published data. We computationally screened various chemical libraries and validated the senolytic action of ginkgetin, periplocin and oleandrin in human cell lines under various modalities of senescence. The compounds have potency comparable to known senolytics, and we show that oleandrin has improved potency over its target as compared to best-in-class alternatives. Our approach led to several hundred-fold reduction in drug screening costs and demonstrates that artificial intelligence can take maximum advantage of small and heterogeneous drug screening data, paving the way for new open science approaches to early-stage drug discovery.
134Ginkgetin Alleviates Inflammation and Senescence by Targeting STING.PubMed
Yadan Liu, Jialin Ye, Zisheng Fan, et al.
Adv Sci (Weinh). 2025 Jan;12(2):e2407222. doi: 10.1002/advs.202407222. Epub 2024 Nov 19.
Ginkgo biloba extract is reported to have therapeutic effects on aging-related disorders. However, the specific component responsible for this biological function and its mechanism of action remain largely unknown. This study finds that Ginkgetin, an active ingredient of Ginkgo biloba extract, can alleviate cellular senescence and improve pathologies in multiple tissues of aging mice. To reveal the molecular mechanism of Ginkgetin's anti-aging effect, a graph convolutional network-based drug "on-target" pathway prediction algorithm for prediction is employed. The results indicate that the cGAS-STING pathway may be a potential target for Ginkgetin. Subsequent cell biological and biophysical data confirmed that Ginkgetin directly binds to the carboxy-terminal domain of STING protein, thereby inhibiting STING activation and signal transduction. Furthermore, in vivo pharmacodynamic data showed that Ginkgetin effectively alleviates systemic inflammation in Trex1 mice and inhibits the abnormally activated STING signaling in aging mouse model. In summary, this study, utilizing an artificial intelligence algorithm combined with pharmacological methods, confirms STING serves as a critical target for Ginkgetin in alleviating inflammation and senescence. Importantly, this study elucidates the specific component and molecular mechanism underlying the anti-aging effect of Ginkgo biloba extract, providing a robust theoretical basis for its therapeutic use.
135Identification of drug combinations on the basis of machine learning to maximize anti-aging effects.PubMed
Sun Kyung Kim, Peter C Goughnour, Eui Jin Lee, et al.
PLoS One. 2021 Jan 28;16(1):e0246106. doi: 10.1371/journal.pone.0246106. eCollection 2021.
Aging is a multifactorial process that involves numerous genetic changes, so identifying anti-aging agents is quite challenging. Age-associated genetic factors must be better understood to search appropriately for anti-aging agents. We utilized an aging-related gene expression pattern-trained machine learning system that can implement reversible changes in aging by linking combinatory drugs. In silico gene expression pattern-based drug repositioning strategies, such as connectivity map, have been developed as a method for unique drug discovery. However, these strategies have limitations such as lists that differ for input and drug-inducing genes or constraints to compare experimental cell lines to target diseases. To address this issue and improve the prediction success rate, we modified the original version of expression profiles with a stepwise-filtered method. We utilized a machine learning system called deep-neural network (DNN). Here we report that combinational drug pairs using differential expressed genes (DEG) had a more enhanced anti-aging effect compared with single independent treatments on leukemia cells. This study shows potential drug combinations to retard the effects of aging with higher efficacy using innovative machine learning techniques.
136A novel functional peptide, named EQ-9 (ESETRILLQ), identified by virtual screening from regenerative cell secretome and its potential anti-aging and restoration effects in topical applications.PubMed
Wang Feifei, Su Wenrou, Kang Sining, et al.
Peptides. 2023 Nov;169:171078. doi: 10.1016/j.peptides.2023.171078. Epub 2023 Aug 12.
Skin aging refers to a degenerative process that can be affected and regulated by intrinsic and extrinsic factors. The mesenchymal stem cell secretome covers a considerable number of regenerative molecules with anti-aging effects in a wide variety of circumstances. However, it is complex, time-consuming, and costly to identify specific compounds from thousands of natural molecules using conventional methods. With the development of computational biology and machine learning, an efficient workflow was generated to identify novel peptides with anti-aging and skin restoration potential. One of the candidate peptides was discovered and subsequently truncated to a novel peptide named EQ-9, with promising anti-aging effects for topical applications at a concentration of 10 ppm validated by experimental validation. The above-described paradigm is expected to be further applied to the virtual screening of novel peptide molecules targeting specific biological functions from a wide variety of natural resources.
137AI-powered therapeutic target discovery.PubMed
Frank W Pun, Ivan V Ozerov, Alex Zhavoronkov
Trends Pharmacol Sci. 2023 Sep;44(9):561-572. doi: 10.1016/j.tips.2023.06.010. Epub 2023 Jul 19.
Disease modeling and target identification are the most crucial initial steps in drug discovery, and influence the probability of success at every step of drug development. Traditional target identification is a time-consuming process that takes years to decades and usually starts in an academic setting. Given its advantages of analyzing large datasets and intricate biological networks, artificial intelligence (AI) is playing a growing role in modern drug target identification. We review recent advances in target discovery, focusing on breakthroughs in AI-driven therapeutic target exploration. We also discuss the importance of striking a balance between novelty and confidence in target selection. An increasing number of AI-identified targets are being validated through experiments and several AI-derived drugs are entering clinical trials; we highlight current limitations and potential pathways for moving forward.
138Integrating Machine Learning with Multi-Omics Technologies in Geroscience: Towards Personalized Medicine.PubMed
Nikolaos Theodorakis, Georgios Feretzakis, Lazaros Tzelves, et al.
J Pers Med. 2024 Aug 31;14(9):931. doi: 10.3390/jpm14090931.
Aging is a fundamental biological process characterized by a progressive decline in physiological functions and an increased susceptibility to diseases. Understanding aging at the molecular level is crucial for developing interventions that could delay or reverse its effects. This review explores the integration of machine learning (ML) with multi-omics technologies-including genomics, transcriptomics, epigenomics, proteomics, and metabolomics-in studying the molecular hallmarks of aging to develop personalized medicine interventions. These hallmarks include genomic instability, telomere attrition, epigenetic alterations, loss of proteostasis, disabled macroautophagy, deregulated nutrient sensing, mitochondrial dysfunction, cellular senescence, stem cell exhaustion, altered intercellular communication, chronic inflammation, and dysbiosis. Using ML to analyze big and complex datasets helps uncover detailed molecular interactions and pathways that play a role in aging. The advances of ML can facilitate the discovery of biomarkers and therapeutic targets, offering insights into personalized anti-aging strategies. With these developments, the future points toward a better understanding of the aging process, aiming ultimately to promote healthy aging and extend life expectancy.
139ExplaiNAble BioLogical Age (ENABL Age): an artificial intelligence framework for interpretable biological age.PubMed
Wei Qiu, Hugh Chen, Matt Kaeberlein, et al.
Lancet Healthy Longev. 2023 Dec;4(12):e711-e723. doi: 10.1016/S2666-7568(23)00189-7. Epub 2023 Nov 6.
BACKGROUND: Biological age is a measure of health that offers insights into ageing. The existing age clocks, although valuable, often trade off accuracy and interpretability. We introduce ExplaiNAble BioLogical Age (ENABL Age), a computational framework that combines machine-learning models with explainable artificial intelligence (XAI) methods to accurately estimate biological age with individualised explanations. METHODS: To construct the ENABL Age clock, we first predicted an age-related outcome (eg, all-cause or cause-specific mortality), and then rescaled these predictions to estimate biological age, using UK Biobank and National Health and Nutrition Examination Survey (NHANES) datasets. We adapted existing XAI methods to decompose individual ENABL Ages into contributing risk factors. For broad accessibility, we developed two versions: ENABL Age-L, based on blood tests, and ENABL Age-Q, based on questionnaire characteristics. Finally, we validated diverse ageing mechanisms captured by each ENABL Age clock through genome-wide association studies (GWAS) association analyses. FINDINGS: Our ENABL Age clock was significantly correlated with chronological age (r=0·7867, p<0·0001 for UK Biobank; r=0·7126, p<0·0001 for NHANES). These clocks distinguish individuals who are healthy (ie, their ENABL Age is lower than their chronological age) from those who are unhealthy (ie, their ENABL Age is higher than their chronological age), predicting mortality more effectively than existing clocks. Groups of individuals who were unhealthy showed approximately three to 12 times higher log hazard ratio than healthy groups, as per ENABL Age. The clocks achieved high mortality prediction power with an area under the receiver operating characteristic curve of 0·8179 for 5-year mortality and 0·8115 for 10-year mortality on the UK Biobank dataset, and 0·8935 for 5-year mortality and 0·9107 for 10-year mortality on the NHANES dataset. The individualised explanations that revealed the contribution of specific characteristics to ENABL Age provided insights into the important characteristics for ageing. An association analysis with risk factors and ageing-related morbidities and GWAS results on ENABL Age clocks trained on different mortality causes showed that each clock captures distinct ageing mechanisms. INTERPRETATION: ENABL Age brings an important leap forward in the application of XAI for interpreting biological age clocks. ENABL Age also carries substantial potential in practical settings, assisting medical professionals in untangling the complexity of ageing mechanisms, and potentially becoming a valuable tool in informed clinical decision-making processes. FUNDING: National Science Foundation and National Institutes of Health.
140An evaluation of aging measures: from biomarkers to clocks.PubMed
Qingyi Wang, Tongyao Hou, Qiwen Wang, et al.
Biogerontology. 2023 Jun;24(3):303-328. doi: 10.1007/s10522-022-09997-4. Epub 2022 Nov 23.
With the increasing number of aged population and growing burden of healthy aging demands, a rational standard for evaluation aging is in urgent need. The advancement of medical testing technology and the prospering of artificial intelligence make it possible to evaluate the biological status of aging from a more comprehensive view. In this review, we introduced common aging biomarkers and concluded several famous aging clocks. Aging biomarkers reflect changes in the organism at a molecular or cellular level over time while aging clocks tend to be more of a generalization of the overall state of the organism. We expect to construct a framework for aging evaluation measurement from both micro and macro perspectives. Especially, population-specific aging clocks and multi-omics aging clocks may better fit the demands to evaluate aging in a comprehensive and multidimensional manner and make a detailed classification to represent different aging rates at tissue/organ levels. This framework will promisingly provide a crucial basis for disease diagnosis and intervention assessment in geroscience.
141Sarcopenia and Cardiogeriatrics: The Links Between Skeletal Muscle Decline and Cardiovascular Aging.PubMed
Dimitrios Anagnostou, Nikolaos Theodorakis, Christos Hitas, et al.
Nutrients. 2025 Jan 14;17(2):282. doi: 10.3390/nu17020282.
Sarcopenia, an age-related decline in skeletal muscle mass, strength, and function, is increasingly recognized as a significant condition in the aging population, particularly among those with cardiovascular diseases (CVD). This review provides a comprehensive synthesis of the interplay between sarcopenia and cardiogeriatrics, emphasizing shared mechanisms such as chronic low-grade inflammation (inflammaging), hormonal dysregulation, oxidative stress, and physical inactivity. Despite advancements in diagnostic frameworks, such as the EWGSOP2 and AWGS definitions, variability in criteria and assessment methods continues to challenge standardization. Key diagnostic tools include dual-energy X-ray absorptiometry (DXA) and bioimpedance analysis (BIA) for muscle mass, alongside functional measures such as grip strength and gait speed. The review highlights the bidirectional relationship between sarcopenia and cardiovascular conditions such as heart failure, aortic stenosis, and atherosclerotic cardiovascular disease, which exacerbate each other through complex pathophysiological mechanisms. Emerging therapeutic strategies targeting the mTOR pathway, NAD+ metabolism, and senescence-related processes offer promise in mitigating sarcopenia's progression. Additionally, integrated interventions combining resistance training, nutritional optimization, and novel anti-aging therapies hold significant potential for improving outcomes. This paper underscores critical gaps in the evidence, including the need for longitudinal studies to establish causality and the validation of advanced therapeutic approaches in clinical settings. Future research should leverage multi-omics technologies and machine learning to identify biomarkers and personalize interventions. Addressing these challenges is essential to reducing sarcopenia's burden and enhancing the quality of life for elderly individuals with comorbid cardiovascular conditions. This synthesis aims to guide future research and promote effective, individualized management strategies.
142XAI-enabled neural network analysis of metabolite spatial distributions.PubMed
Wenwu Ma, Lanfang Luo, Kun Liang, et al.
Anal Bioanal Chem. 2023 Jun;415(14):2819-2830. doi: 10.1007/s00216-023-04694-8. Epub 2023 Apr 21.
We used deep neural networks to process the mass spectrometry imaging (MSI) data of mouse muscle (young vs aged) and human cancer (tumor vs normal adjacent) tissues, with the aim of using explainable artificial intelligence (XAI) methods to rapidly identify biomarkers that can distinguish different classes of tissues, from several thousands of metabolite features. We also modified classic neural network architectures to construct a deep convolutional neural network that is more suitable for processing high-dimensional MSI data directly, instead of using dimension reduction techniques, and compared it to seven other machine learning analysis methods' performance in classification accuracy. After ascertaining the superiority of Channel-ResNet10, we used a novel channel selection-based XAI method to identify the key metabolite features that were responsible for its learning accuracy. These key metabolite biomarkers were then processed using MetaboAnalyst for pathway enrichment mapping. We found that Channel-ResNet10 was superior to seven other machine learning methods for MSI analysis, reaching > 98% accuracy in muscle aging and colorectal cancer datasets. We also used a novel channel selection-based XAI method to find that in young and aged muscle tissues, the differentially distributed metabolite biomarkers were especially enriched in the propanoate metabolism pathway, suggesting it as a novel target pathway for anti-aging therapy.
144Detailing the biomedical aspects of geroscience by molecular data and large-scale "deep" bioinformatics analyses.PubMed
Andreas Simm, Anne Großkopf, Georg Fuellen
Z Gerontol Geriatr. 2024 Aug;57(5):355-360. doi: 10.1007/s00391-024-02329-w. Epub 2024 Aug 1.
As scientists investigated the molecular mechanisms of the biology of aging, they discovered that these are malleable and can enhance healthy longevity by intervening in the drivers of aging, which are leading to disease, dysfunction and death. These exciting observations gave birth to the field of geroscience. As the mechanisms of aging affect almost all mechanisms of life, detailed molecular mechanistic knowledge must be gained or expanded by considering and integrating as many types of data as possible, from genes and transcripts to socioenvironmental factors. Such a large-scale integration of large amounts of data will in turn profit from "deep" bioinformatics analyses that provide insights beyond contextualizing and interpreting the data in the light of knowledge from databases such as the Gene Ontology. The authors suggest that "deep" bioinformatics, employing methods based on artificial intelligence, will be a key ingredient of future analyses.