Soutenance de thèse : Fang WAN

Multi-criteria optimization for the management of intensive care beds in an epidemic context

Doctorante : Fang WAN

Laboratoire INSA : DISP

Ecole doctorale : ED512 Infomaths

In this study, we explore the various challenges of healthcare resource management in the context of the COVID-19 pandemic, such as multiple uncertainties, multi-objective considerations, and diverse research subjects. The focus is on issues like medical resource allocation, priority setting, predictive models, and hyperparameter optimization. First, we conduct an in-depth investigation into intensive care units (ICU) bed allocation strategies, primarily considering how to reasonably distribute ICU beds among different patient types to maximize admission rates, patient satisfaction, and resource utilization, while maintaining scheduling stability under the influence of uncertainties. Then, Considering the upstream resources related to ICUs, especially the direct impact of operating room (OR) on ICU bed allocation, we coordinated the optimization of the OR and ICU modules. We examined the impact of OR allocation on ICU bed distribution, particularly in the context of random emergency patient arrivals. We explored how to minimize peak ICU bed demand, reduce delays in elective surgeries, and minimize healthcare staff overtime. Next, due to the random arrival of emergency patients, uncertain surgery duration, and prolonged length of stays (LOS) of patients in ICU, we first predicted emergency patient arrivals, patients' surgery durations and LOS, and then allocated ORs for both emergency and elective patients. Our goal is to provide timely emergency services and improve the utilization of ORs and ICU beds while minimizing the cancellation of elective surgeries. We also compared the performance of large language model (LLM) and traditional algorithms in the allocation of ORs and ICU beds. The performance of the models was further improved through hyperparameter optimization. Finally, we summarize the key findings of this research and provide suggestions for future research directions, particularly in improving the resilience and adaptability of medical resource management systems.