Demand analysis and capacity management for hospital emergencies using advanced forecasting models and stochastic simulation

•According to the results presented in this paper for three hospitals, it is possible to perform demand forecasting with great confidence for hospital emergency services. After testing several forecasting methods, SVR provided best results in terms of variance and accuracy. Based on this forecasting, a logic for managing capacity was developed for one hospital. Such logic uses the comparison between the forecasted demand and the available medical resources and a simulation model to assess the performance of different configurations of facilities and resources. These analyses provide hospital managers a decision tool for determining the number and distribution of medical resources on the emergency service, based on a cost/benefit analysis of resources and service improvement. The results above support the task of assigning doctors to different kinds of boxes, defining their work schedules, and considering additional doctors..•The forecasting method and the capacity management logic proposed in this paper have been validated and accepted by hospital managers and staff, and are currently in use in the Hospital Luis Calvo Mackenna (HLCM) which is a major pediatric hospital in Santiago, Chile. For this use, there was a need for formal processes, which embed the forecasting model and the resources management logic, including a support computing system. The results with the implemented processes have been encouraging and the National Health Authorities are considering extending the whole design concept to other public hospitals in Chile.•It is important to notice that the design of the processes, with the embedded analytics and IT support, is not a one-time effort. Its design includes the periodical execution and adaptation of the processes under changing conditions, such as unexpected demand, for example, epidemic episodes and new campaigns, which require adapting capacity.•An interesting feature of the work reported in this paper is the integration of several methods presented independently in different publications. Such methods are forecasting, simulation, processes design, and IT support. The methods’ integration facilitates the practical use of quantitative models, since, when their use is independent and on a one-time basis, they may produce interesting results, but there is no guarantee of a practical impact. The integration takes care explicitly of designing a solution for routine use, which also has adaptation capabilities to facilitate use under changi