View article

A discrete event simulation (DES) model with Auto-Regressive Integrated Moving Average (ARIMA) forecast inputs, sampled service times, resource capacities and scheduled resource changes was used to project inpatient populations, referral waitlists, and bed utilization for a five-site hospital system with over 10,000 patients. Based on a SAS Simulation Studio platform, the model can project arbitrary subpopulations on a three-year horizon and perform “what if” experiments with bed allocations and patient flows. This application demonstrates the utility of DES for providers with statutory obligations to serve forensic populations, while also exposing the limitations presented by missing data, non-random variations in data collection across sites, and sizable exogenous variation.