面向II型糖尿病人群的健康管理蕴藏着巨大的社会价值，针对精准给药的智能化决策与控制是慢病管理的核心技术。从医学知识角度出发，本研究通过对多源医疗文本的信息挖掘，构建涵盖医学常识和临床经验的医学知识图谱，并设计个性化体征与医学知识图谱相结合的状态图谱生成算法，在医学知识指导下实现对患者体征的精准刻画。从药物机理角度出发，采用药代动力学-药效动力学理论测量用药前后患者体征的变化情况，并根据其真实状态反馈，客观地评价给药方法的有效性，同时为优化给药行为提供科学依据。最后，将集成阶段性研究成果，提出一种基于强化学习的给药方法优化模式，形成给药方案制定、状态图谱演变、药效量化评估三者协同的动态调节过程，实现人体状态与药物作用持续交互的闭环控制系统。通过模拟医生的专业知识体系，该系统能够根据患者个性化体征制定具有靶向性的给药方案，有效提高医疗资源的利用率、改善慢病人群的医疗服务水平。

Health management for people with type 2 diabetes has huge social value. Intelligent decisions and control methods of precision medication are the key technologies for chronic disease management. From the perspective of medical knowledge, this study employs text mining technology to construct a medical knowledge graph by multi-source medical texts that cover medical common sense and clinical experience. To depict patient portraits under the guidance of medical knowledge, we design an algorithm for generating the state graph, which combines personalized signs and medical knowledge graph. From the perspective of drug mechanism，it is planned to adopt the pharmacokinetic-pharmacodynamic theory to measure the changes of the patient's physical signs before and after medication, and to objectively evaluate the effectiveness of the medication strategy based on the state feedback of patient. Meanwhile, it provides a scientific basis for the adjustment of medication strategy. Finally, the periodic research will be integrated to propose an optimization model of drug delivery strategy based on reinforcement learning, which forms a dynamic regulation process coordinating with the formulation of drug delivery strategy, the evolution of state graph, and evaluation of drug efficacy. The optimization model aims to achieve a closed-loop control system for continuous interaction between the human system and drug action. By simulating the knowledge schema of professional doctors, the control system can formulate targeted medication plans according to personalized signs, thereby effectively improving the utilization of medical resources and the medical service level of the chronic illness.