基于机器学习的外周动脉疾病不良事件研究进展

Research progress on adverse events of peripheral arterial disease based on machine learning

摘要:

外周动脉疾病(PAD)是一类由动脉粥样硬化引起的、影响全身外周动脉的高负担疾病,其不良事件(截肢、再干预、死亡等)的精准风险分层对临床决策至关重要。机器学习模型为PAD患者的个性化风险预测提供了新手段,可弥补传统统计方法难以捕捉复杂非线性关系的不足。文章系统梳理了逻辑回归、XGBoost、随机森林、神经网络等机器学习方法在PAD不良事件风险预测中的应用,对比了不同模型的预测性能与适用场景。并进一步探讨了模型解释性、数据隐私与伦理等关键挑战,旨在为PAD不良事件的精准风险预测提供方法学参考,并为未来研究指明方向。

Abstract:

Peripheral arterial disease (PAD) is a kind of high burden disease caused by atherosclerosis and affecting the peripheral arteries of the whole body. The precise risk stratification of adverse events (amputation, reintervention, death, etc.) is crucial to clinical decision-making. Machine learning models provide a new means for personalized risk prediction of PAD patients, which can compensate for the shortcomings of traditional statistical methods in capturing complex nonlinear relationships. This article systematically reviews the application of machine learning methods including Logistic regression, XGBoost, random forest and neural networks in the risk prediction of adverse events of PAD, and the predictive performance and applicable scenarios of different models were compared. And further explored key challenges such as model interpretability, data privacy, and ethics, aiming to provide methodological references for accurate risk prediction of adverse events in PAD and indicate directions for future research.