面向多检查的门诊患者调度研究

发布时间：2018-08-25 16:51

【摘要】：及时的检查对于患者病情诊断和治疗非常重要。然而,患者不同的紧急程度、检查项目的多样性、以及患者的行为因素如失约等,使门诊患者调度问题难以求解。为解决该问题,本文考虑患者对两种检查项目的不同需求,两类不同的紧急程度,以及患者失约和医生加班,建立了有限时域马尔可夫决策过程(MDP)模型,目标是使得患者检查所得的期望收益最大化以及期望加班时间惩罚成本最小化。由于MDP模型复杂,难以用解析方法来分析最优控制策略,因此本文基于MDP模型进行数值实验,观察最优解的结构特征,进一步构造了两种参数化启发式调度策略,并采用遗传算法对调度策略的参数进行优化。数值实验比较了最优控制策略、两种启发式调度策略以及先到先服务规则,实验结果表明本文所提的调度策略性能偏离最优解不超过10%;当工作负荷非常大时,启发式调度策略远远优于先到先服务规则。
[Abstract]:Timely examination is very important for the diagnosis and treatment of the patient's condition. However, the different emergency degree of patients, the diversity of examination items, and the behavioral factors of patients, such as failure, make it difficult to solve the problem of outpatient scheduling. In order to solve this problem, a finite time domain Markov decision process (MDP) model is established in this paper, considering the different needs of patients for two kinds of examination items, two kinds of different emergency levels, and the patient's failure and doctors' overtime. The goal is to maximize the expected return from patient testing and minimize the penalty cost of expected overtime. Because of the complexity of MDP model, it is difficult to analyze the optimal control strategy by analytic method. Therefore, based on the numerical experiment of MDP model, the structural characteristics of the optimal solution are observed, and two parameterized heuristic scheduling strategies are further constructed. Genetic algorithm is used to optimize the parameters of scheduling policy. Numerical experiments have compared the optimal control strategy, two heuristic scheduling strategies and the first-come first served rule. The experimental results show that the proposed scheduling policy performance deviates from the optimal solution by less than 10 percent, and when the workload is very heavy, The heuristic scheduling strategy is much better than the first come first served rule.
【作者单位】：上海交通大学工业工程与管理系;
【基金】：国家自然科学基金资助项目(71471113)
【分类号】：O225;R197.3

【相似文献】