Real-Time Patient Transfer under Ambulance Offload Delay

Problem Definition: Ambulance offload delays (AOD) occur when emergency medical services (EMS) transfer a patient to a busy hospital emergency department (ED), and the patients cannot be immediately transferred from the paramedics to the hospital staff. Relevance: It impacts both the patient and the EMS system, preventing timely patient care and keeping the ambulance from returning to service new calls. Methodology: In this paper, we formulate the patient transfer problem with AOD as a Markov decision process (MDP) and develop an approximate dynamic programming (ADP) approach to overcome the curse of dimensionality. We formulate the problem based on post-decision states and derive two solution algorithms using the approximate policy iteration method with domain knowledge. Results: Numerical results show that the transfer policy obtained from our solution method consistently outperforms the benchmark static policy by 49.44%(±6.57%). We apply our method to St. Paul, Minnesota system using one year of historical data. The results showed a reduction in total weighted wait time by 48.33%(±12.24%). Managerial Impli- cations: Our findings suggest that the proposed approach can effectively mitigate AOD problems, improve quality of care, and increase overall system efficiency