Forecasting ED Overcrowding With Statistical Methods: A Prospective Validation Study

Emergencies

emergency department, overcrowding, forecasting, artificial intelligence, deep learning

The aim of this study is to prospectively validate statistical forecasting tools that have been widely used retrospectively in forecasting ED overcrowding

Emergency department (ED) overcrowding is a chronic international issue that has been repeatedly associated with detrimental treatment outcomes such increased 10-day-mortality. Forecasting future overcrowding would enable pre-emptive staffing decisions that could alleviate or prevent overcrowding along with its detrimental effects. Over the years, several predictive algorithms have been proposed ranging from generalized linear models to state space models and, more recently, deep learning algorithms. However, the performance of these algorithms has only been reported retrospectively and the clinically significant accuracy of these algorithms remains unclear. In this study the investigators aim to investigate the accuracy of the previously reported ED forecasting algorithms in a prospective setting analogous to the way these tools would be used if used implemented as a decision-support system in a real-life clinical setting.

No related papers found yet.