基于数据驱动面向家庭医疗的高血压并发症风险评估与多资源协同实时调度优化

Risk evaluation for hypertension complications and real-time scheduling optimization of medical resources that devoted to grasping the “golden time” for rescue by providing prompt medical treatment for patients are crucial for promoting the health management of patients with hypertension (high blood pressure), preventing patient deterioration and saving patients' lives. Aiming at solving the problems of risk evaluation for hypertension complications and the multi-resource collaborative real-time scheduling optimization that is committed to ensuring the critical time window for rescuing in home health care, this project adopts the bayesian inference with hybrid intelligent algorithm optimization to evaluate the risks of hypertension complications based on the factor analysis of the suddenness and randomness of hypertension complications, the uncertainty of medical service process and the multi-resource collaborative sharing, and the data-driven risk identification of hypertension complications, so as to provide guidance for the prevention and intervention of hypertension complications. On this basis, this project explores the collaborative allocation mechanism of multi-resources in home health care by considering the actual constraints such as the emergency degree of patients with hypertension complications, emergency medical services response time, the limited key medical resources and the continuity of medical care, and then adopts the data-driven simulation methods based on stochastic optimization to study the multi-resource collaborative real-time scheduling optimization. Research results are expected to provide the decision-making basis decision makers on real-time scheduling optimization. This project extends the data-driven risk identification and risk evaluation methods for hypertension complications, and provides some new insights for the multi-resource collaborative real-time scheduling in home health care.

高血压并发症的风险评估与抢夺“救护黄金期”的医疗资源的实时调度优化对于高血压患者健康管理、抑制病情恶化乃至挽救患者生命意义重大。针对面向家庭医疗的高血压并发症风险评估以及保障抢救时间窗的多资源协同实时调度优化的难题，在探讨高血压并发症的突发性和随机性、医疗服务过程的不确定性、多资源共享性等因素的基础上，研究基于数据驱动的高血压并发症的风险识别，采用混合智能算法优化的贝叶斯推理对高血压并发症的风险进行评估，从而为高血压并发症的预防和干预提供指导。在此基础上，综合考虑并发症患者的紧急程度、服务响应时间、关键医疗资源的有限性、服务的持续性等实际约束，探讨家庭医疗多资源协同分派机制，进而采用基于随机优化和数据驱动的仿真方法，研究多资源协同实时调度优化，以便为决策人员提供实时调度优化的决策依据。项目研究拓展了基于数据驱动的高血压并发症风险识别和评估方法，可为家庭医疗的多资源协同实时调度提供新思路。

高血压并发症风险评估与医疗资源实时调度优化对高血压患者健康管理、寿命延长以及“救护黄金期”保障和患者生命挽救意义重大。针对面向家庭医疗的高血压并发症风险评估以及抢救时间窗强化的多资源协同实时调度优化难题，项目分为五个专题进行研究，取得了系列重要研究成果。（1）基于数据驱动面向家庭医疗的高血压并发症风险识别。针对高血压并发症的突发性和随机性特征，项目提出混合文本挖掘和专家知识的方法识别了高血压并发症高危风险因素。（2）基于数据驱动面向家庭医疗的高血压并发症风险评估。针对高血压并发症如脑卒中、心肌梗塞等疾病风险预测问题，采用深度学习、机器学习等人工智能技术建立预测准确率提高和泛化性能加强的高血压并发症风险预测模型，有助于解决“早发现早治疗”的现实医疗问题。（3）基于数据驱动面向家庭医疗的多资源协同分派机制。根据社区内和社区间资源协同的不同特性及面临的不同难题，研究社区内和社区间的关键医疗资源协同调度问题，提出适应度高的调度方法及高效求解算法。（4）基于数据驱动面向家庭医疗的高血压并发症多资源协同实时调度。考虑紧急患者插入和临时取消服务两种常见突发事件，重点探究突发事件下的医疗人员实时调度优化问题。（5）构建基于数据驱动面向家庭医疗的高血压并发症风险评估与实时调度优化决策平台。结合上海市多家医院和社区医疗数据，进行了仿真实验、案例验证和综合实证研究。项目提出了一套基于数据驱动面向家庭医疗的高血压并发症风险评估和多资源协同实时调度理论与方法，可为家庭医疗运作和医疗决策提供理论指导。项目研究成果转化应用于指导上海市多家社区家庭医疗运作管理实践，取得显著社会效益，并形成了一定的示范效应。