AI’s Role in Easing ER Overcrowding

On August 12, 2025, a study published in the Mayo Clinic Proceedings: Digital Health revealed that artificial intelligence (AI) could play a significant role in managing emergency room overcrowding. The research indicates that AI programs can help medical staff predict which ER patients are likely to require hospital admission, doing so several hours earlier than traditional methods.

AI Outperforms Human Predictions

The study involved training an AI program on nearly two million patient visits, demonstrating a slight edge over ER nurses in predictive accuracy. Researchers found that while nurses achieved about 81% accuracy in predicting admissions, the AI’s prediction accuracy rose to 85%. Jonathan Nover, the study’s lead researcher and vice president of nursing and emergency services at Mount Sinai Health System, noted the urgency of addressing the escalating crisis of overcrowding and boarding in emergency departments nationwide.

The Crisis of ER Overcrowding

ER overcrowding and patient boarding have become critical national issues, often leading to prolonged wait times and worsening patient outcomes. According to recent studies, up to 35% of patients requiring admission can wait four or more hours in spare rooms or hallways. During peak periods, nearly 5% may wait an entire day for a bed. This reality underscores the necessity for streamlined admission processes.

AI as a Solution

Nover likens the situation in emergency departments to industries like airlines and hotels, which effectively utilize booking systems to forecast demand. He emphasized the absence of such a system in healthcare, expressing hope that AI could serve as a form of ‘reservation’ for patient admissions.

The AI tool developed during this project aims to forecast admissions needs even before formal orders are placed, thereby offering hospitals critical insights to enhance patient flow and outcomes.

Training the AI

The AI’s training focused on data from over 1.8 million ER visits from 2019 to 2023, allowing it to identify patterns to better anticipate admissions. In the study, this machine learning model was evaluated alongside over 500 ER nurses, highlighting the complementary roles of AI and human expertise in clinical settings.

The Future of AI in Emergency Care

Looking forward, the research team plans to implement their AI tool into real-time workflows, carefully monitoring its impact on boarding times and overall patient flow in emergency departments. As Robert Freeman, the chief digital transformation officer at Mount Sinai, remarked, this advancement is not intended to replace clinicians, but to support them, ultimately leading to better patient care through timely admissions.

Freeman emphasized the real-world applicability of AI in health care, stating that it is becoming a viable solution shaped by the professionals on the front lines of care.

Learn More

The American College of Emergency Physicians offers additional resources on the issues of ER boarding and overcrowding.