Answers are grounded in this paper and cite their sources.
In the United States, emergency department congestion stealthily undermines the safety net of public hospitals for uninsured, underinsured, and medically complex populations, and the use of forecasting tools is limited where it is desired most. This research article proposes and tests a set of predictive modeling techniques for predicting the volume of patients in the emergency department (ED) and downstream hospital resource use for the next day or multiple days within public hospitals in the United States, based on 4 years of anonymized encounter data and calendar, weather, epidemiology, and hospital operational data. Different models, including naive, seasonal autoregressive integrated moving average with exogenous regressors (SARIMAX), Prophet, random forest, extreme gradient boosting (XGBoost), long short-term memory (LSTM), and attention-based transformer architectures were trained and evaluated using 1-day, 7-day, and 14-day forecasting horizons, and the results were converted to a discrete-event simulation of staffing and bed management.
No takes yet. Share an insight, caveat, or question.
Gidisu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: