Ambulance Location Models for Emergency Medical Services: A Survey with a Focus on Mexico and Developing Countries

Abstract: Emergency Medical Service (EMS) systems face critical challenges inambulance location and dispatching decisions, challenges that are exacerbated in developing countries where resources are severely constrained. Although the literature on EMS location models is extensive, existing surveys largely reflect the assumptions of well-resourced systems and pay limited attention to settings where complete emergency coverage is unattainable. The contribution of this survey is threefold. First, we review the evolution of EMS optimization models from classical deterministic formulations such as the location set covering and maximal covering location models, through probabilistic queuing-based approaches, to stochastic programming models that explicitly handle demand uncertainty, emphasizing the modeling features most relevant to resource-constrained environments, such as partial coverage and heterogeneous fleets. Second, we focus on Mexico as a representative developing-country context, where ambulance availability often falls 30–60% below World Health Organization recommendations, and we present a critical comparative framework contrasting two recent real-world applications: a deterministic multi-period double standard model applied to Tijuana’s Red Cross operations, and a two-stage stochastic maximum expected coverage model with partial coverage through decay functions for heterogeneous fleets in Monterrey. This framework compares the models across uncertainty treatment, coverage philosophy, fleet heterogeneity, location–dispatch integration, and computational performance, and yields practical guidance for model selection. Third, we identify research directions specific to developing-country EMS systems, including multi-provider coordination, data limitations, and hospital capacity integration. Our analysis shows that meaningful improvements are achievable through formulations that explicitly acknowledge resource scarcity rather than adapting frameworks designed for well-resourced systems.