PanOpt: A Nationwide Joint Optimization of Dynamic Bed Allocation and Patient Transfer in Pandemics

Dynamic bed allocation and patient admission have emerged as critical strategies for hospitals during pandemics, driving research toward minimizing costs while effectively managing hospital resources. Existing studies have explored bed management in the intensive care unit, resource allocation, and surge capacity planning. However, gaps remain in addressing inter-hospital patient transfers. This paper addresses the pressing challenges hospitals face in resource management during the rapid spread of infectious diseases, which increase the risk of cross-infection and financial strain on healthcare systems. To tackle these issues, we introduce a nationwide joint optimization framework of dynamic bed allocation and patient transfer in pandemics, which we refer to as PanOpt. PanOpt presents a novel integer linear programming model for optimizing patient allocation to specific wards based on case severity, dynamic bed allocation across wards, and inter-hospital patient transfers based on capacity. Our proposed framework minimizes total healthcare system costs while reducing cross-infection risks and ensuring efficient utilization of resources. PanOpt is evaluated under various system parameters and benchmarked against two models. In the first model, inter-hospital transfers are prohibited, whereas in the second model, reallocation and inter-hospital transfers are prohibited. Results show that PanOpt outperforms existing techniques, significantly lowering operational costs and providing a robust solution for healthcare resource optimization.