农村留守老人健康问题严峻，潜在健康风险较高，已成为当前老年健康管理领域及卫生政策制定过程中亟待解决的重要问题。目前有关农村留守老人健康风险的研究，缺乏从个体、微观环境、宏观环境三层面综合分析农村留守老人生理、心理、社会三维健康影响因素，尚未建立全面科学的农村留守老人健康风险评估模型和健康风险测评工具。本研究以农村留守老人为研究对象，采用荟萃分析法等总结农村留守老人健康影响因素；运用决策树从个 体、微观环境和宏观环境三层面分析各分支点影响因素的分类及所属；采用关联规则GRI算法挖掘农村留守老人生理、心理、社会三维健康风险因子；运用BP人工神经网络构建农村留守老人健康风险评估模型，并采用ROC曲线等对模型进行验证；结合ASP开发平台研发实用有效的农村留守老人健康风险测评工具，为早期发现农村留守老人健康问题及制定相关健康策略提供依据，从而提高农村留守老人的健康水平和生命质量。

The rural left-behind elderly’s severe health problems and high potential health risks, have become an important issue of the current elderly health management in health policy-making process. Current researches on rural left-behind elderly health risks lacked from individual, micro environment and macro environment three levels to comprehensively analyze the physical, psychological and social health factors of rural left-behind elderly, and not yet established a comprehensive scientific rural left-behind elderly health risk assessment model and effective health risk assessment instrument. This study will select rural left-behind elderly as the research object, use meta-analysis to summarize health factors; Apply decision tree to classifying each branch point influencing factors from the individual, micro environment and macro environment three levels; Use GRI association rule to excavate physical, psychological, social three-dimension health risk factors of rural left-behind elderly; Utilize BP artificial neural network to build rural left-behind elderly health risk assessment model, and verify the model with ROC curves; Combine with ASP platform to develop interactive, dynamic and efficient rural left-behind elderly health risk assessment instrument, providing a basis for health problems detecting and health policy making of rural left-behind elderly, in order to improve the health level and quality of life of rural left-behind elderly.