AI and the World's Hard Problems · Medicine & drug discovery · arXiv · 2026-10-23
Do Machine Learning-Guided Proactive Full Capacity Protocols Mitigate Emergency Department Overcrowding? A Simulation-Based Evaluation
The study tests whether machine learning can trigger Full Capacity Protocols in emergency departments before crowding builds, rather than after thresholds are exceeded. In a simulation calibrated on a large US hospital, a model predicting waiting-room crowding six hours ahead drove proactive activation, compared with reactive activation over 365 days and 10 replications. Proactive activation significantly reduced the time above the crowding threshold and protocol duration, and reduced crowding variability and extreme congestion, while length-of-stay and wait-time differences were not statistically significant.