中医临床预测模型研究现状及方法进展OACSTPCD
Research status and method advances of clinical prediction models for traditional Chinese medicine
临床预测模型是借助现有临床数据构建合适的统计模型,并总结特定临床场景下某结局事件发生概率的规律性.临床预测模型在中医药领域的应用较少.针对中医药领域预测模型研究的制作流程进行全面介绍,包括:变量筛选(分为基于回归分析和基于机器学习两大类)、模型构建(线性回归、logistic回归、Cox回归和机器学习算法中的随机森林算法)、模型评价(AUC、一致性指数、Hosmer-Lemeshow拟合优度检验、校准曲线、DCA决策曲线、NRI与IDI)、模型验证(内部验证和外部验证)及模型的可视化展示,以期为该方法在中医药领域应用的可行性和前景提供参考.
Clinical prediction models involve the construction of appropriate statistical models using existing clinical data to summarize the probability patterns of specific clinical outcomes.However,their application in the field of traditional Chinese medicine(TCM)remains limited.This paper comprehensively introduces the production process of predictive model research in TCM,covering variable selection(classified into regression analysis and machine learning-based approaches),model construction(including linear regression,logistic regression,Cox regression,and machine learning algorithms such as random forest),model evaluation(metrics like AUC,consistency index,Hosmer-Lemeshow goodness of fit test,calibration curve,DCA decision curve,NRI,and IDI),model validation(internal and external validation),and model visualization.It aims to provide insights into the feasibility and potential of applying prediction models in the TCM field.
施泽阳;王禹毅;孙源;卜志军;刘龙坤;刘建平;刘兆兰
北京中医药大学循证医学中心 北京 100029北京中医药大学循证医学中心 北京 100029||重庆市中医院北京中医药大学循证医学中心 北京 100029||烟台市疾病预防控制中心北京中医药大学中医学院
基础医学
中医预测模型变量筛选模型构建模型评价模型验证
TCMprediction modelvariable selectionmodel constructionmodel evaluationmodel validation
《现代中医临床》 2024 (003)
21-27 / 7
国家自然科学基金项目(No.82374298);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)
评论