Verma Nipun, Choudhury Ashok, Singh Virendra, Duseja Ajay, Al-Mahtab Manum, Devarbhavi Harshad, Eapen Chundamannil E, Goel Ashish, Ning Qin, Duan Zhongping, Hamid Saeed, Jafri Wasim, Butt Amna Shubhan, Shukla Akash, Tan Soek-Siam, Kim Dong Joon, Hu Jinhua, Sood Ajit, Goel Omesh, Midha Vandana, Ghaznian Hashmik, Sahu Manoj Kumar, Lee Guan Huei, Treeprasertsuk Sombat, Shah Samir, Lesmana Laurentius A, Lesmana Rinaldi C, Prasad V G Mohan, Sarin Shiv K
Department of Hepatology, Postgraduate Institute of Medical Education and Research, Chandigarh, India.
Department of Hepatology, Institute of Liver and Biliary Sciences, New Delhi, India.
Liver Int. 2023 Feb;43(2):442-451. doi: 10.1111/liv.15361. Epub 2022 Oct 11.
We hypothesized that artificial intelligence (AI) models are more precise than standard models for predicting outcomes in acute-on-chronic liver failure (ACLF).
We recruited ACLF patients between 2009 and 2020 from APASL-ACLF Research Consortium (AARC). Their clinical data, investigations and organ involvement were serially noted for 90-days and utilized for AI modelling. Data were split randomly into train and validation sets. Multiple AI models, MELD and AARC-Model, were created/optimized on train set. Outcome prediction abilities were evaluated on validation sets through area under the curve (AUC), accuracy, sensitivity, specificity and class precision.
Among 2481 ACLF patients, 1501 in train set and 980 in validation set, the extreme gradient boost-cross-validated model (XGB-CV) demonstrated the highest AUC in train (0.999), validation (0.907) and overall sets (0.976) for predicting 30-day outcomes. The AUC and accuracy of the XGB-CV model (%Δ) were 7.0% and 6.9% higher than the standard day-7 AARC model (p < .001) and 12.8% and 10.6% higher than the day 7 MELD for 30-day predictions in validation set (p < .001). The XGB model had the highest AUC for 7- and 90-day predictions as well (p < .001). Day-7 creatinine, international normalized ratio (INR), circulatory failure, leucocyte count and day-4 sepsis were top features determining the 30-day outcomes. A simple decision tree incorporating creatinine, INR and circulatory failure was able to classify patients into high (90%), intermediate (60%) and low risk (~20%) of mortality. A web-based AARC-AI model was developed and validated twice with optimal performance for 30-day predictions.
The performance of the AARC-AI model exceeds the standard models for outcome predictions in ACLF. An AI-based decision tree can reliably undertake severity-based stratification of patients for timely interventions.
我们假设人工智能（AI）模型在预测慢加急性肝衰竭（ACLF）的预后方面比标准模型更精确。
我们从亚太肝脏研究学会慢加急性肝衰竭研究联盟（AARC）招募了2009年至2020年期间的ACLF患者。连续90天记录他们的临床数据、检查结果和器官受累情况，并用于AI建模。数据被随机分为训练集和验证集。在训练集上创建/优化了多个AI模型、终末期肝病模型（MELD）和AARC模型。通过曲线下面积（AUC）、准确性、敏感性、特异性和类别精度在验证集上评估预后预测能力。
在2481例ACLF患者中，训练集有1501例，验证集有980例，极端梯度提升交叉验证模型（XGB - CV）在预测30天预后的训练集（0.999）、验证集（0.907）和总体集（0.976）中表现出最高的AUC。在验证集中，XGB - CV模型预测30天预后的AUC和准确性（%Δ）分别比标准的第7天AARC模型高7.0%和6.9%（p < 0.001），比第7天的MELD高12.8%和10.6%（p < 0.001）。XGB模型在预测7天和90天预后时也具有最高的AUC（p < 0.001）。第7天的肌酐、国际标准化比值（INR）、循环衰竭、白细胞计数和第4天的脓毒症是决定30天预后的主要特征。一个包含肌酐、INR和循环衰竭的简单决策树能够将患者分为高死亡风险（约90%）组、中死亡风险（约60%）组和低死亡风险（约20%）组。开发了一个基于网络的AARC - AI模型，并进行了两次验证，其对30天预后的预测具有最佳性能。
AARC - AI模型在预测ACLF预后方面的性能超过了标准模型。基于AI的决策树能够可靠地对患者进行基于严重程度的分层，以便及时进行干预。
