Multi-stage stochastic programming for dynamic nurse scheduling and bed allocation with buffer wards
针对大型医院在疫情中设立缓冲病房的场景，提出多阶段随机规划模型，动态调整护士排班和床位分配，以降低交叉感染率并减轻护士生理负担。
Large public hospitals have instituted the dedicated epidemic buffer wards to effectively register and screen patients, thereby preventing nosocomial cross-infection in the event of an outbreak. Due to the high uncertainty in the demands of different types of patients and the physiological burden on nurses across different types of wards, it becomes crucial to dynamically adjust the nurse scheduling and bed allocation. A multi-stage stochastic programming model in this paper is developed that integrates nurse scheduling and bed allocation to efficiently utilise limited and necessary healthcare resources with epidemic buffer wards. A data-driven decision-making framework using the rolling horizon procedure is proposed to address the model. Extensive numerical experiments are performed to demonstrate the effectiveness of the optimal decision framework and to illustrate how sensitive the outcomes are to variations in key parameters. The experiments show that the proposed optimal decision framework for bed and nurse scheduling plays a crucial role in significantly reducing cross-infection rates within hospitals during epidemic outbreaks. Furthermore, the framework contributes to alleviating the physiological burden on nurses, thereby enhancing their capacity to provide care under stressful conditions.