【作者】 周杰；

【导师】 李军；

【作者基本信息】 西南交通大学 ， 资源优化管理， 2016， 博士

【摘要】 当前,人民群众对医疗服务的需求快速增加。需求快速增加而医疗资源总体有限导致医患之间的供需矛盾日益突出。患者“就医体验差”现象在大型医院普遍存在。要缓解这一矛盾,需要加大对医疗卫生事业的投入力度、推进分级诊疗体系的实施、提升医疗服务行业的运作效率。根据世界卫生报告(2010)的保守估计,约20%—40%的卫生资源因效率低下而被浪费。作为医疗服务的具体实施者,各级医疗机构可以从微观层面解决日常运作效率不高的问题。医疗资源运作管理的核心问题之一即是医疗资源的容量管理问题。预约调度是医疗机构进行容量管理的有效途径,它提前将医疗资源容量进行分配,它能够提高资源的利用效率、匹配供需关系、平缓病人流量,同时降低病人的等待时间。根据实地走访S医院发现放射科CT室是该院的一个服务瓶颈,粗放的容量管理方式降低了人员设备利用效率,加重了病人预约请求往后积压的现象。CT室是先预约后服务的服务系统,且服务系统面临两类独立的病人群体,以先到先服务为原则的常规病人和有非强占优先权的急诊病人。本文以具有上述特征的大型医院CT室作为研究对象,结合CT室业务流程现状,收集相关费用信息和记录数据,在考虑急诊非强占优先权的前提下,基于不同研究目标,采用不同研究方法对现行容量管理策略进行改进。首先,分别从预约规则设计和排队规则设计视角分析常规病人等待时间。在预约规则设计中,从为急诊预留时隙的数量和放置位置两个决策变量出发,形成不同的容量管理策略,研究各策略对常规病人期望等待时间的影响。建立随机计算模型刻画等待时间。比较分析表明,预留时隙集中放置在服务期中期能显著减少常规病人等待时间。从排队规则设计中,分别建立等待制M/F1/1排队系统和混合制Geom/NB/1排队系统。通过解析分析发现,无论是等待制还是混合制排队系统,等待队列中处于相同相位的常规病人的等待时间一致,但任一常规病人的等待时间不同。通过案例分析给出了混合制排队系统中常规病人损失率趋于0的病人到达强度和最大容量限制参数组合。其次,分别从预约规则设计和排队规则视角对服务系统的总成本进行分析。在预约规则设计中,预留时隙集中放置在后期能使医疗资源空闲时间和加班时间最小。医患双方加权总成本随预留时隙数量增加先减少后增加。将适当数量的预留时隙集中放置在中期能有效减少医患双方加权总成本。在等待制排队系统中,构建一个凸规划,通过Kuhn-Tucker条件得到常规病人的最优到达强度来权衡常规病人的等待时间和医疗资源空闲时间,实现病人等待时间和医疗资源空闲时间加权和最小。最后,在精细划分常规病人费用信息的前提下,以服务系统期望总收益最大化为目标建立马尔科夫决策模型描述CT室容量管理问题,体现常规病人延迟检查的心理感知与等待时间的非线性关系。运用收益管理容量控制理论,通过边际分析技巧获得各等级病人的预约限制数量,以此保证CT室接受预约请求的增量效益大于0。当不考虑空余时隙时,由模型的结构性质诱导出若干容量管理策略改善服务系统的期望总收益。当考虑空余时隙时,在确定空余时隙的数量和位置后,分两步得到的容量管理策略也可以改善服务系统的期望总收益。仿真结果显示,两种途径得到的容量管理策略均可在医院的加和收益、常规病人的直接等待成本、预约请求被推迟的惩罚成本、医疗资源加班成本和空闲成本之间达到很好的平衡。本论文针对不同目标效用给出了预约操作建议。本论文的研究成果可应用到具有相似特征的医疗资源容量管理问题中去。更多还原

【Abstract】 Healthcare industry faces increasing demand and lengthy waiting time for a wide scope of medical service. The long waits and overcrowded in the large-size hospital makes patients uneasy. Several measures could be employed to solve these problems, such as increasing healthcare industry financial input, carrying out the hierarchical diagnosis system, promoting the utilization of the medical resources. About 20%-40% of the medical resources are wasted according to the conservative estimation of World Health Report (2010). Healthcare managers and policy makers face pressure to manage the medical resources more efficiently and effectively. The core of healthcare operations management is capacity management. Appointment scheduling system is an efficiency lever to manage medical capacity to provide timely access to health services. Appointment scheduling allocates the capacity in advance to effectively utilize the resource, smooth the flow of demand and decrease the waiting time of patients.The radiology department is a key department within the hospital after interviewing the medical staff and administrator in S hospital. CT scan is a bottleneck for providing patients timely service. The backlog of patients and medical resources’ idle time frequently occurred due to the unreasonable capacity management policy. CT department is an appointment-based service system. There are two independent sources of patients: first-come-first-serve regular patient and non-preemptive priority emergency patient.This work focuses on the CT service system which has the above characteristics. The capacity management policies are improved by different methods based on the operation procedure and record data. One assumption that emergency patients have non-preemptive priorities is made throughout this work.Firstly, the direct waiting time of regular patient is considered from appointment rule design and queuing discipline design, respectively. In the appointment rule design, we consider the impact of two decision variables (the quantity of the free slots reserved for emergency patients and the position of these slots) on the expected direct waiting time of regular patients. The waiting time is calculated by mathematical model. Numerical comparisons show that allocating the free slots in the middle of the service session could dramatically decrease the direct waiting time of regular patients. In the queuing discipline design, we construct the M/Ek/1 queueing system with waiting and Geom/NB/1 mixed queueing system, respectively. The analytical results show that the waiting time of the same phase patient in these two queueing systems are the same. However, the waiting time of an arbitrary patient in these two queueing system are different. We also give the values of the patient flow and the upper limit for regular patient so as to let the limit value of loss probability of regular patients be zero in the mixed system.Secondly, the total cost of the service system is considered from appointment rule design and queuing discipline design, respectively. In the appointment rule design, placing the free slots in the end of the session performs best on idle time and overtime of medical resources. The total cost of the service system decreases firstly and then increases as the increases of the number of reserved slots. Allocating the free slots in the middle of the session could effectively reduce the expected total cost. In the queuing discipline design, we construct a nonlinear convex optimization problem which includes the penalty cost of unutilized time of the device and the waiting cost of the regular patients. The optimal arrival rate of regular patients is given by Kuhn-Tucker condition so as to minimize the total cost.Lastly, the regular patients are divided into several classes based on the summed up revenues. Our model describes the direct waiting time of regular patient under emergency random arrival as well as nonlinear relationship between the direct waiting time and the perception of the waiting time. The appointment problem is described as a Markov finite horizon dynamical programming with the goal of improving the total expected reward. Drawing upon the ideas from capacity control theory of revenue management, the booking limit numbers for each patient class are obtained by marginal analysis to ensure the incremental benefit of receiving an appointment is positive. If there is no free slot, some heuristic policies are obtained to improve the total expected reward of hospital. If there are free slots, the booking limit numbers are obtained by two steps marginal analysis. The numerical analysis under different scenarios shows that the total expected reward are improved. It shows that our capacity management policies could achieve trade-off between the summed up revenues for hospital, the direct waiting cost for regular patients, the penalty cost for delaying regular requests, overtime cost for medical resources and penalty cost for idling medical resources.Different appointment scheduling policies for different objective utility functions are given in this work. We expect that this study could apply to the capacity management of the similar medical resources.更多还原