医技资源调度与病人排程马尔科夫决策优化模型研究

With the explosion and aging of the popularity, the public healthcare systems worldwide are facing the escalating healthcare cost, increasing demand and limited resources. One of the most critical and challenging problems that the Grade-III Class-A hospitals in China are confronting is, the overload of large-scale medical diagnostic equipment, and that patients need to make the appointment in advance and wait in long queue for diagnostic services. This project develops Markov Decision Processes (MDP) optimization models for an effective and efficient resource allocation and patient scheduling for diagnostic facilities, in order to decrease patients' waiting time and increase the utilization of the facilities. It consists of two topics (models): (a) appointment scheduling and dynamic resource allocation for multi-segment patients and multi-diagnostic facilities; (b) managing resource allocation between emergency patients and elective patients under stochastic service time. Model (a) focuses on outpatient scheduling and dynamic resource allocation among inpatients, outpatients and emergency patients through the finite-horizon MDP. It has combined advanced scheduling and allocation scheduling effectively. Model (b) focuses on resource allocation between emergency and elective patients under stochastic service time. It has taken into consideration the double layer of uncertainty from demand and supply. We conduct theoretical analysis on the optimal, sub-optimal and heuristic policies and design efficient algorithms. The research results of this project and its applications can, not only help reduce healthcare cost and improve service efficiency, but also provide important theory and methodology for the hospital managers.

随着人口膨胀与老龄化，全球的公共医疗卫生系统都面临着费用增长、需求激增、资源稀缺等问题。本项目针对中国城市三级甲等医院医技诊断设备超负荷运转、病人接受检查服务需预约、排队、等候问题，通过对病人排程和医技资源的调度建立马尔科夫决策优化模型，以减少病人等待时间、提高医技资源利用效率。本项目包括两个专题和模型：(a) 多类病人、多台医疗诊断设备的预约排程和资源分配；(b) 随机不确定服务时间下紧急病人和非紧急病人医技资源调度。模型(a)针对医技资源在门诊、住院和急诊不同部门病人之间的调度排程，有效结合预约排程与分配排程；模型(b)针对服务时间不确定下医技资源在紧急和非紧急病人间的实时调度，有效解决需求与资源总能量的双重不确定性。本研究从科学量化的角度对医技资源和病人排程进行研究，其最优、启发式策略的理论分析、算法设计及实际应用，可降低医疗成本、提高服务效率，为医院管理者提供重要的理论和方法。

随着人口膨胀与老龄化，全球的公共医疗卫生系统都面临着费用增长、需求激增、资源稀缺等问题。本项目针对中国城市三级甲等医院医技诊断需求远高于供给且极具波动、而管理低效，病人接受检查服务需预约、排队、等候问题，通过对医技资源调度和病人排程建立优化模型，以提高医技资源的社会效益和经济效益。本项目开展了以下几个方面的研究：(1) 多类病人多台医疗诊断设备的预约策略问题；(2) 对一类期初决策、期间动态调整的马尔科夫决策过程模型优化；(3) 通过离散事件仿真对随机不确定服务时间下紧急病人和非紧急病人医技资源调度排程问题；(4) 通过仿真研究医技资源专用与共享问题。本课题从建模和方法的理论研究上取得了一定成果并能很好地运用于实际，通过马尔科夫决策优化模型和离散事件计算机仿真对医技资源的调度排程进行建模，并用医院实际数据进行拟合和实验，我们的研究成果给医院在考虑非预约急诊病人随机到达情况下对门诊、住院病人的预约问题提供了简单易行、性能优良可靠的解决方案，降低医疗成本、提高服务效率和效益，为医院管理者提供了重要的理论和方法。