School of Biological Science and Medical Engineering, Beihang University, Beijing, CHINA.
Department of Critical Care Medicine, Chinese PLA General Hospital, Beijing, CHINA.
Med Sci Sports Exerc. 2021 Sep 1;53(9):1826-1834. doi: 10.1249/MSS.0000000000002674.
Rhabdomyolysis (RM) is a complex set of clinical syndromes that involves the rapid dissolution of skeletal muscles. Mortality from RM is approximately 10%. This study aimed to develop an interpretable and generalizable model for early mortality prediction in RM patients.
Retrospective analyses were performed on two electronic medical record databases: the eICU Collaborative Research Database and the Medical Information Mart for Intensive Care III database. We extracted data from the first 24 h after patient ICU admission. Data from the two data sets were merged for further analysis. The merged data sets were randomly divided, with 70% used for training and 30% for validation. We used the machine learning model extreme gradient boosting (XGBoost) with the Shapley additive explanation method to conduct early and interpretable predictions of patient mortality. Five typical evaluation indexes were adopted to develop a generalizable model.
In total, 938 patients with RM were eligible for this analysis. The area under the receiver operating characteristic curve (AUC) of the XGBoost model in predicting hospital mortality was 0.871, the sensitivity was 0.885, the specificity was 0.816, the accuracy was 0.915, and the F1 score was 0.624. The XGBoost model performance was superior to that of other models (logistic regression, AUC = 0.862; support vector machine, AUC = 0.843; random forest, AUC = 0.825; and naive Bayesian, AUC = 0.805) and clinical scores (Sequential Organ Failure Assessment, AUC = 0.747; Acute Physiology Score III, AUC = 0.721).
Although the XGBoost model is still not great from an absolute performance perspective, it provides better predictive performance than other models for estimating the mortality of patients with RM based on patient characteristics in the first 24 h of admission to the ICU.
横纹肌溶解症（RM）是一组涉及骨骼肌迅速溶解的复杂临床综合征。RM 的死亡率约为 10%。本研究旨在开发一种可解释且可推广的 RM 患者早期死亡率预测模型。
对两个电子病历数据库（eICU 协作研究数据库和医疗信息集市强化护理 III 数据库）进行回顾性分析。我们从患者 ICU 入院后 24 小时内提取数据。对两个数据集的数据进行合并，进一步分析。合并后的数据集随机分为两部分，70%用于训练，30%用于验证。我们使用机器学习模型极端梯度提升（XGBoost）和 Shapley 加法解释方法对患者死亡率进行早期和可解释的预测。采用了 5 种典型的评价指标来开发一个可推广的模型。
共有 938 名 RM 患者符合本分析条件。XGBoost 模型预测医院死亡率的受试者工作特征曲线（ROC）下面积（AUC）为 0.871，敏感性为 0.885，特异性为 0.816，准确性为 0.915，F1 评分为 0.624。XGBoost 模型的性能优于其他模型（逻辑回归，AUC=0.862；支持向量机，AUC=0.843；随机森林，AUC=0.825；朴素贝叶斯，AUC=0.805）和临床评分（序贯器官衰竭评估，AUC=0.747；急性生理学评分 III，AUC=0.721）。
尽管从绝对性能的角度来看，XGBoost 模型并不是很好，但它基于 ICU 入院后 24 小时内患者的特征，为估计 RM 患者的死亡率提供了比其他模型更好的预测性能。
