患者就医决策与患者流优化分布建模分析研究

Although national health condition of China has improved greatly along with the growing GDP, "inaccessibility" is still criticized as a major problem of the healthcare delivery system in China. The patient flow distribution currently in China is irrational. In this proposal, we take healthcare service quality, price and the waiting time as the main factors that will affect the patients' decision on choose the healthcare facilities. Also these factors will be considered as incentive to improve the patient flow distribution among different level's healthcare facilities. By dynamically and recurrently assign different level incentive, along with the analysis of patients' behavior, we will construct different utility function to different patients, and finally get the Bayesian Nash equilibrium of patients' decision under incomplete information. In order to model the patient congestion and limited healthcare resource, we propose to establish Markov model with blocking and batch service size. The blocking probability in general hospitals, the waiting time of patients to get healthcare service and the utilization of healthcare facilities can be obtained from the model. The advantage of this project is that we propose a novel optimization tool to improve the patient flow in China, which takes the patients decision making, the impact of government incentive, and healthcare facilities performance into account. These will support the government to have an quantitative tool to improve the patient flow, and help to optimize the healthcare resource allocation finally.

针对当前我国"看病难"，患者流分布不合理的现状，本研究将综合考虑影响患者选择医疗机构进行就诊的三大因素：医疗服务质量、就诊价格和就医等待时间，将其作为政府的激励因素，基于数学建模和量化分析合理优化患者流分布。通过循环优化设置不同的激励程度，结合患者就医行为效用分析，得到患者选择不同医疗机构的均衡决策，即不同医疗机构患者流到达的分布情况。为量化评估该患者流分布对于政策实施及整个医疗系统的影响效果，进一步建立带堵塞和批处理的马尔科夫模型，模拟当前医疗系统就诊号数量有限的现实情况，模型求解获得不同医疗机构就诊的等待时间和各类医疗机构的服务率，以此来说明激励政策所带来的量化效果。本研究将提出一套考虑患者决策、政府激励政策实施效果和医疗服务机构运作方式三者动态过程及其相互影响的多因素、循环优化决策的分析工具，以期为政府提供量化的决策支持方法，研究成果将有助于我国医疗改革解决"看病难"问题。

本研究首先从宏观层面对区域内不同级别医疗机构之间患者流分布情况进行随机优化建模研究。结合效用函数、博弈论对患者就医行为建模，通过证明不完全信息下多人博弈的均衡存在，设计求解算法，获取了基于效用最大化的患者就医选择行为预测。通过价格调控和质量改善的激励方式，考虑了激励因素对患者就医选择行为的量化影响程度。针对我国分级医疗系统的结构特征，进一步研究带转诊的医院之间和医院内部的多层患者流分布，通过仿真优化，决策出多目标优化下的最优分层医疗系统患者的首诊率，双向转诊率。结合海淀区的实证数据，研究优化后得到的最优患者流分布能增加系统服务能力23%，缩短近30%的综合医院首诊患者系统逗留时间，社区卫生服务中心资源利用率增幅高达238%。考虑门诊预约系统是患者获取医疗服务的初始环节，本研究进一步从运营层面，依据我国当前“挂号难”的现状，通过考虑患者取消行为下、带堵塞和批服务特征的门诊预约系统进行数学模型构建和求解，制定不同供需情况下的优化门诊预约策略。本研究所提出的优化预约规则可以降低医疗需求不确定性带来的影响，为门诊的有效、高效运作提供有力保障。综上，本研究提取医疗服务系统的典型特征和主要矛盾进行了模型构建，并结合实证数据进行了量化分析，推进和发展了运营管理在医疗服务系统的研究工作，为我国的医疗服务改革提供了优化理论依据。