A Stochastic Programming Approach for Ambulance Dynamic Relocation

Pre-hospital Emergency Medical Service (EMS) provides the critical function of on-site medical treatment, stabilization of patients and transportation. In general, the performance of EMS can be measured by the Time of Arrival at Hospital (TAH), defined as the time interval from the dispatch of an ambulance until the arrival of the patient at a hospital. A better management of ambulances can reduce the TAH by taking into consideration of short-term demand forecast. This study aims to propose a dynamic relocation system to integrate EMS demand forecasting and with an ambulance deployment model to decrease the TAH. In our work, we develop a dynamic system, a stochastic spatial temporal network allocation model, and a Lagrangian dual decomposition with branch and bound algorithm to optimize the ambulance operations. By simultaneously considering the forecasting demands and ambulance immediate deployment, we can obtain a more optimized deployment plan for the next time interval. The purpose of the model is to assist close-to-real-time ambulance relocation decisions. The results show that the system and approaches we propose have the potential to enhance the performance of pre-hospital EMS and could be implemented in the real world.