面向健康管理的个体健康画像概念模型构建  

Research on the Construction of Individual Health Portrait Conceptual Model for Health Management

作　　者：张馨月 牟冬梅[1,2] 王长聪 王萍 Zhang Xinyue;Mu Dongmei;Wang Changcong;Wang Ping(School of Public Health,Jilin University,Changchun 130021,China;Clinical Research Department,The First Hospital of Jilin University,Changchun 130021,China)

机构地区：[1]吉林大学公共卫生学院,吉林长春130021 [2]吉林大学第一医院,吉林长春130021

出　　处：《现代情报》2025年第5期152-166,共15页Journal of Modern Information

基　　金：国家自然科学基金项目“‘知识-数据’双轮驱动多模态电子病历数据组织模式与应用研究”(项目编号:72374081);吉林省科技发展计划项目“面向健康自主管理的个人诊疗信息平台开发与应用”(项目编号:20240304164SF)。

摘　　要：[目的/意义]面向健康管理场景,以真实世界个体健康医疗数据为研究对象,构建多维度的个体健康画像概念模型,提升个体健康画像的实用性。[方法/过程]在对现有个体健康画像研究评述的基础上,面向健康管理应用场景,通过分析真实世界个体和群体健康数据,提出符合真实世界数据条件下多维度健康画像的标签体系,构建包括人口特征、健康素养、行为习惯及健康状态4个维度的健康画像概念模型,并以英国生物样本库中50万脂肪肝患者建立健康画像,依据该健康画像进行预测研究。[结果/结论]针对真实世界数据构建4个维度健康画像,画像包括的多标签涵盖了患者的基础指标、客观指标和可控行为,反映个体健康状态。通过精准的健康状态辨识和远期结局评估,展示了从单维度到多维度3种模型不同时间点的ROC曲线。结果表明,随着4种维度信息的逐步纳入,AUC值从0.58逐渐增加到0.64,验证了“4P”模式下多维度健康画像在预测远期健康结果方面的有效性,证明了个体健康画像能够帮助医生和个体更好地理解健康状态,促进精准化健康管理的实施。[Purpose/Significance]Aiming at health management scenarios,taking real-world individual health and medical data as the research object,we construct a multi-dimensional individual health portrait conceptual model to improve the practicality of individual health portraits.[Method/Process]Based on the review of existing research on individual health portraits,aiming at health management application scenarios,by analyzing real-world individual and group health data,a label system for multi-dimensional health portraits that conformed to real-world data conditions was proposed,and a conceptual model of health portraits including four dimensions:demographic characteristics,health literacy,behavioral habits,and health status was constructed.A health portrait was established for 500000 patients with fatty liver disease in the UK biobank,and a predictive study was conducted based on the health portrait.[Result/Conclusion]A four-dimensional health portrait is constructed based on real-world data.The multiple labels included in the portrait cover the patient's basic indicators,objective indicators and controllable behaviors,reflecting the individual's health status.Through accurate health status identification and long-term outcome assessment,the ROC curves of three models at different time points from single dimension to multi-dimensional were demonstrated.The results showed that with the gradual inclusion of information from the four dimensions,the AUC value gradually increased from 0.58 to 0.64,verifying the effectiveness of multi-dimensional health portraits under the“4P”model in predicting long-term health outcomes,and proving that individual health portraits can help doctors and individuals better understand their health status and promote the implementation of precision health management.

关 键 词：用户画像 健康画像 画像维度 健康管理 真实世界数据 

分 类 号：R058[医药卫生] TP311[自动化与计算机技术—计算机软件与理论]