患者不守时情形下数据驱动的医院门诊预约调度方法

This project mainly focuses on the research of patient unpunctuality that addressing the practical operation management needs of hospital outpatient departments. Based on realistic unpunctuality data of patient, the data driven method is applied to analyze the behavior characteristics and subdivide the categories of patient unpunctuality. The outpatient manager would make a reasonable appointment schedule according to the characteristics of different patient unpunctuality categories, aiming at increasing the utilization of outpatient medical resources and reducing the doctor’s idle time and patients’ waiting times. This project extends the outpatient appointment scheduling with single stage and homogeneous patients to the scheduling considering two stages and multiple categories of patient unpunctual arrival behavior, by designing the service rule for patients in different categories, proposing the modeling method of the out-of-order arrivals of patients, formulating the stochastic programming models for the outpatient appointment scheduling system considering multiple uncertain factors and designing the effective optimization algorithm for the programming models from the perspective of analytical optimization. The researches contents include clustering and modeling probability distribution function of patient unpunctuality in data driven approach; outpatient appointment scheduling optimization considering different patient unpunctuality categories with single stage service; outpatient appointment scheduling optimization considering return-visits and different patient unpunctuality categories with two stage service. The research results would not only enrich the theory and methods of outpatient appointment scheduling considering unpunctuality, but also provide technical supports and management suggestions for the outpatient management practice.

本项目以医院门诊运作管理为背景，将患者不守时到达情形作为重点研究对象，旨在基于现实的患者不守时行为数据，运用数据驱动方法分析患者的不守时行为特征、细分患者不守时到达类型，根据患者不守时类型特征合理调度患者，提高门诊医疗资源利用率、减少不必要的医生空闲和患者等待时间。将原有单阶段、考虑单一不守时类型患者的预约调度问题扩展到数据驱动的两阶段、考虑多种不守时类型患者的预约调度问题，从解析优化视角设计多种不守时类型患者的服务规则，提出患者在失序到达情况下的建模方法，构建综合考虑多种不确定因素的数学规划模型并设计优化算法。研究内容包括：数据驱动的患者不守时类型聚类和概率分布函数拟合；考虑多种患者不守时类型的单阶段门诊预约调度方法；考虑回诊环节和多种不守时类型的两阶段门诊预约调度方法。研究成果不仅可以丰富不守时情形下门诊预约调度问题的理论与方法，也为医疗门诊的管理实践提供技术支持和政策建议。

本项目以门诊预约患者的不守时行为作为研究对象，旨在面向患者不守时行为特征有针对性地优化预约调度系统。基于真实数据，运用数据驱动方法对患者不守时行为进行了聚类和特征分析。针对不守时患者的失序到达难点，以解析优化视角提出患者在失序到达情况下的一般建模方法，设计面向多种不守时类型患者的服务规则，构建综合考虑多种不确定因素的数学规划模型并设计优化算法。研究内容和成果包括：从概率角度对不守时情形的随机性进行描述，建立了不守时行为的到达概率模型，提出不守时行为的理论化建模方法；研究了面向不守时患者的线上序列预约调度问题，理论证明了序列预约患者数量的单峰性和预约时段对于不守时患者的非最优调度条件；研究了考虑多种不守时类型的患者预约调度优化问题，建立患者虚拟到达模型，提出了基于邻域搜索的强化学习算法；进一步针对多服务台场景，研究了面向多医生的不守时患者预约分配与调度问题，设计了两阶段迭代优化算法。本项目为不守时情形下医院门诊的运作管理提供了有效的理论、方法和技术支持。