糖尿病严重威胁我国1.1亿多居民健康，支出占我国医疗全部费用的13%，每年并发症致死达130万人。大数据智能研究应用是推进糖尿病健康的预测性、预防性、个性化、参与性（医疗健康领域称为P4）管理技术和水平的关键。本项目针对如下核心问题展开研究：（1）建立原始数据、知识发现、预测决策三层体系框架，从整体解决糖尿病智能健康管理及多源异构大数据融合在实现P4目标所面临的医疗健康特殊需求和数据冲突、高维、冗余、时序和稀疏性等问题；（2）建立时序因果关系发现模型与代价敏感多视角模型提升糖尿病并发症预测和预防的精准性，解决医疗领域根据因果关系判断和正误诊疗代价不平衡的特殊需求；（3）建立知识发现与智能决策集成框架提升糖尿病管理的个性化和自动化智能管理水平。项目实现的三层体系框架及整体技术创新可望有效推进糖尿病P4管理技术和水平，为医疗健康相关领域的智能管理奠定关键技术基础。

Diabetes mellitus (DM) seriously threatens the health of over 110 million people in China. The annual DM related expenditure accounts for 13% of the total medical expenses in our country, and more than 1.3 million people die from DM complications each year. Big data-driven intelligent research and application are the key to advance P4 (i.e., Predictive, Preventive, Personalized, and Participatory management) health management of diabetes mellitus. This project aims at addressing the following core research challenges: (1) establishing a three-tiered systematic framework including raw data, knowledge discovery and decision making to comprehensively solve the domain-specific problems and data-processing challenges such as data conflict, high dimensionality, redundancy, and feature sparse problems, which are introduced by DM intelligence management and multi-source heterogeneous data fusion in pursuit of P4 objectives; (2) improving the accuracy of prediction and prevention of diabetic complications through development of temporal causal relationship discovery model and cost sensitive multi-view model, addressing the physicians’ need for causal relationships and cost sensitive decision support in healthcare; (3) improving the personalized and automated diabetes management by establishing an action mining-based intelligent framework integrated with knowledge discovery. In this project, the three-tier systematic framework and overall technological innovation would effectively promote the P4 intelligent management of DM, and would lay a key technical foundation for health management.