Hospital selection problem for emergency medical services: a data-driven metamodeling and simulation-based optimization approach

Metamodeling-Based Simulation Optimization (MBSO) is a hybrid framework integrating simulation, metamodeling, and optimization to address complex decision-making challenges. While MBSO has demonstrated efficacy across various sectors, its application within healthcare systems remains underutilized. This study evaluates the implementation of MBSO for the hospital selection problem in Fez, Morocco. A Discrete Event Simulation (DES) model was developed to represent patient flow and clinical processes, providing the synthetic data required to train 11 machine learning-based metamodels. Following our experimental results, the five top-performing metamodels were selected and subsequently integrated into three of the most widely used metaheuristic algorithms in the literature: Genetic Algorithm (GA), Differential Evolution (DE), and Tabu Search (TS). The results validate the effectiveness of the proposed approach. Specifically, the coupling of CatBoost with both GA and DE yielded the most robust outcomes, outperforming OPTQUEST. These findings highlight the potential of MBSO in healthcare and suggest new pathways for enhancing decision-support systems through advanced hybrid optimization frameworks.