疗效预测模型评价指标及其在中医药领域的应用研究  

Evaluation metrics of therapeutic prediction models and their application in the field of Traditional Chinese Medicine

作　　者：李志强[1] 王彦丁 徐慧琳 王雪峰 苑艺 卜志军 刘兆兰[1] 刘建平[1] LI Zhiqiang;WANG Yanding;XU Huiling;WANG Xuefeng;YUAN Yi;BU Zhijun;LIU Zhaolan;LIU Jianping(Centre for Evidence-Based Chinese Medicine,Beijing University of Chinese Medicine,Beijing 100029;Institute for Medical Information Processing,Biometry,and Epidemiology(IBE),Pettenkofer School of Public Health,Faculty of Medicine,LMU Munich;School of Economics,Anhui University)

机构地区：[1]北京中医药大学循证医学中心,北京100029 [2]路德维希-马克西米利安-慕尼黑大学医学院医学信息处理、生物统计和流行病学研究所,佩滕科费尔公共卫生学院 [3]安徽大学经济学院

出　　处：《现代中医临床》2024年第6期26-31,54,共7页Modern Chinese Clinical Medicine

基　　金：国家自然科学基金项目(No.82374298,No.81830115);北京中医药大学学科后备带头人支持计划(No.90010960920033);国家中医药管理局高水平中医药重点学科建设项目(No.zyyzdxk-2023249)。

摘　　要：在临床决策支持系统中,疗效预测模型起到了关键作用,其主要是通过机器学习算法来揭示预测因素与疗效结局之间的关联。模型的性能评价对于验证其有效性和可靠性至关重要,这对于辅助临床医生制定更精准的诊疗决策,提高治疗效率和患者满意度具有显著影响。不仅可以降低治疗相关的风险和不良反应,还可帮助改善疾病预后。在中医药领域,疗效预测模型及其性能评价指标的应用,展现了提高中医治疗精确性和效果方面的巨大潜力,对促进中医治疗的科学化发展具有重要意义。本文将详细介绍准确率、灵敏度、F1分数、接收者操作特征曲线(ROC曲线)等常用的评价指标,并探讨其在中医领域的应用现状,旨在为医疗工作者提供一个更全面的理解框架,从而为中医临床预测模型性能评价的指标选择提供科学参考。In clinical decision support systems,therapeutic prediction models play a pivotal role,primarily employing machine learning algorithms to reveal the correlations between predictive factors and therapeutic outcomes.Evaluating the performance of these models is crucial to ensure their effectiveness and reliability,which significantly impacts the precision of clinical decision-making by physicians,enhancing treatment efficiency,and patient satisfaction.It not only reduces treatment-related risks and side effects but also improves disease prognosis.In the field of traditional Chinese medicine(TCM),the application of efficacy prediction models and performance evaluation metrics highlights their potential in improving the accuracy and effectiveness of TCM treatments,playing a significant role in enhancing the scientific basis of TCM therapies.This article will provide a detailed introduction to the commonly used evaluation indexes including accuracy,sensitivity,F1 score,ROC curve,etc.,and discuss their current applications in the TCM domain,aiming to offer healthcare professionals a more comprehensive understanding of the framework,so as to serve as a scientific reference for the selection of performance evaluation metrics for TCM clinical prediction models.

关 键 词：疗效预测模型 评价方法 临床试验 中医 

分 类 号：R2-03[医药卫生—中医学]