Department of Pharmacy, School of Applied Sciences, University of Huddersfield, Queensgate Huddersfield HD1 3DH, West Yorkshire, United Kingdom.
Department of Pharmacy, School of Applied Sciences, University of Huddersfield, Queensgate Huddersfield HD1 3DH, West Yorkshire, United Kingdom; School of Biomedical Sciences & Pharmacy, University of Newcastle, Callaghan, Australia.
Int J Med Inform. 2022 Mar;159:104679. doi: 10.1016/j.ijmedinf.2021.104679. Epub 2021 Dec 31.
The advent of clinically adapted machine learning algorithms can solve numerous problems ranging from disease diagnosis and prognosis to therapy recommendations. This systematic review examines the performance of machine learning (ML) algorithms and evaluates the progress made to date towards their implementation in clinical practice.
Systematic searching of databases (PubMed, MEDLINE, Scopus, Google Scholar, Cochrane Library and WHO Covid-19 database) to identify original articles published between January 2011 and October 2021. Studies reporting ML techniques in clinical practice involving humans and ML algorithms with a performance metric were considered.
Of 873 unique articles identified, 36 studies were eligible for inclusion. The XGBoost (extreme gradient boosting) algorithm showed the highest potential for clinical applications (n = 7 studies); this was followed jointly by random forest algorithm, logistic regression, and the support vector machine, respectively (n = 5 studies). Prediction of outcomes (n = 33), in particular Inflammatory diseases (n = 7) received the most attention followed by cancer and neuropsychiatric disorders (n = 5 for each) and Covid-19 (n = 4). Thirty-three out of the thirty-six included studies passed more than 50% of the selected quality assessment criteria in the TRIPOD checklist. In contrast, none of the studies could achieve an ideal overall bias rating of 'low' based on the PROBAST checklist. In contrast, only three studies showed evidence of the deployment of ML algorithm(s) in clinical practice.
ML is potentially a reliable tool for clinical decision support. Although advocated widely in clinical practice, work is still in progress to validate clinically adapted ML algorithms. Improving quality standards, transparency, and interpretability of ML models will further lower the barriers to acceptability.
临床适应的机器学习算法的出现可以解决从疾病诊断和预后到治疗建议等众多问题。本系统评价检查了机器学习（ML）算法的性能，并评估了迄今为止在将其应用于临床实践方面取得的进展。
系统地搜索数据库（PubMed、MEDLINE、Scopus、Google Scholar、Cochrane Library 和 WHO COVID-19 数据库），以确定 2011 年 1 月至 2021 年 10 月期间发表的原始文章。报告涉及人类的临床实践中的 ML 技术和具有性能指标的 ML 算法的研究被认为是合格的。
在 873 篇独特的文章中，有 36 篇符合纳入标准。XGBoost（极端梯度增强）算法显示出最高的临床应用潜力（n=7 项研究）；其次是随机森林算法、逻辑回归和支持向量机，分别（n=5 项研究）。对结果的预测（n=33），特别是炎症性疾病（n=7）受到了最多的关注，其次是癌症和神经精神疾病（n=5）和 COVID-19（n=4）。在所选的 TRIPOD 清单中，36 项纳入研究中有 33 项超过了 50%的质量评估标准。相比之下，根据 PROBAST 清单，没有一项研究能够达到理想的总体偏倚率“低”。相比之下，只有三项研究显示了在临床实践中部署 ML 算法的证据。
ML 是临床决策支持的潜在可靠工具。尽管在临床实践中得到广泛提倡，但仍在努力验证临床适应的 ML 算法。提高 ML 模型的质量标准、透明度和可解释性将进一步降低接受的障碍。
