Objective Chronic disease risk prediction is an important issue in the fields of public health as well as health information technology .Present article reviews the current research situation of chronic disease risk pre‐diction .Three database (PubMedEBSCOEMBASE) was retrieved for the literature collection of chronic disease risk prediction .The chronic disease risk prediction related technology ,method ,and content in three aspects includ‐ed :chronic disease risk factors ,risk prediction model ,and quality assessment of prediction model was discussed . The occurrence and development of chronic diseases are closely related with biological genetic ,living habits ,and social psychological factors .Risk prediction model was based on two methods ,the statistical model and the artificial intelligence model .The quality assessment of chronic disease risk prediction model commonly used global model fit‐ness ,calibration ,receiver operating characteristic (ROC) ,net reclassification improvement (NRI) ,and integrated discrimination improvement (IDI) .In future research ,the public health big data should play more important role , and multi‐factors combined risk prediction model should be best method for risk discrimination .The idea of evi‐dence‐based medicine and ethical issues should give full consideration in chronic risk prediction .%目的:慢性病患病风险预警问题是公共卫生领域的重要课题,也是卫生信息技术领域的研究热点。本文对国外慢性病患病风险预警研究现状进行梳理。采用文献回顾研究方法,检索 PubMed 、EBSCO 、EM‐BASE3个数据库收录的慢性病患病风险预警的相关研究论文,从“慢性病危险因素遴选”、“风险预警模型构建”和“慢性病风险预警模型质量评估”3个方面,对慢性病患病风险预警建模所涉及的技术、方法和内容等问题进行归纳。慢性病的发生、发展与生物遗传、生活习惯、社会心理等因素密切相关;风险预警建模方法可采用基于统计学模型和人工智能的方法;预警模型的质量评估方法一般采用全局模型拟合评估、标度评估、受试者工作特征曲线(ROC)、净再分类改善指数(NRI)和整合辨识力改善指数(IDI)等。在今后的研究中,应该充分发挥公共卫生大数据优势,注重多因素联合作用对慢性病患病风险的影响,遵循循证医学的思想,并充分考虑慢性病患病风险预警中的伦理学问题。
展开▼