基于马尔可夫决策过程的医疗检查匹配预约调度优化

为合理匹配医疗检查任务，提高医生服务质量和医院收益，针对医疗检查即时预约调度问题，考虑医生资质和患者病情差异，以最大化医院收益为目标建立基于马尔可夫决策过程（Markov decision process, MDP）的匹配预约调度模型，并采用逆序值迭代算法对模型进行求解。最后通过仿真实验得到不同资质医生接受不同病情患者的最优预约策略，并与传统先到先服务（first come first service，FCFS）预约策略进行对比分析。结果表明，MDP预约策略是与医生剩余可服务患者数相关的阈值策略，该策略相比于FCFS预约策略所获收益更高，更有利于保障医院稳健运营，且在医疗检查资源越匮乏时MDP预约策略的优势越明显。

Medical Examination Matching Appointment Scheduling Optimization Based on Markov Decision Process

In order to reasonably match the medical examination task and improve the service quality of doctors and the hospital income, according to the instant reservation and scheduling problems of medical treatment, and considering the doctor qualification and the disease differences of patients, the matching reservation scheduling model based on Markov decision process (MDP) was established with the purpose of maximizing the hospital profit, and the inverse value iterative calculation was adopted to solve the model. Finally, through the simulation experiment, the best reservation strategy was obtained that the doctors of different qualifications accepted the patients with different conditions, which was compared and analyzed with the traditional first come first service (FCFS). The result shows that MDP appointment strategy is the threshold strategy relating to the remaining patient quantity served by doctors. This obtained income of this strategy is more than that of FCFS reservation strategy, which is benefit for guaranteeing the steady operation of the hospital. In addition, the advantage of MDP appointment strategy is more obvious when the medical examination resources is lacking.