AI in Emergency Medicine - Promising Improvements and Barriers

Madison Melton - Department of Emergency Medicine

Link: https://doi.org/10.1002/ams2.740

Developments in artificial intelligence (AI) and machine learning (ML) software have shown immense potential in benefiting emergency medicine. These improvements could address common challenges faced in the emergency department (ED) like triage, timely detection of emergent conditions, and long wait times. However, there are also safety concerns about its use in healthcare. 

Current research focuses on triage, medical imaging, and ED operations. Triage is the process in which patients are assessed and prioritized by severity when they first arrive in the ED. It can be a lengthy process and is traditionally done by a specially-trained nurse. The use of ML models here could help expedite triage and predict urgency. Another use is in medical imaging. ED providers do not always have quick access to radiological interpretation, yet timely, accurate readings are key to avoiding misdiagnosis and treatment delay. ML models for medical imaging could help providers promptly identify conditions, and “recent reports suggest that the quality of AI interpretation is not inferior to an expert radiologist” (Mueller 2022). AI could also improve overcrowding, a major issue in the ED, by predicting volume and wait times as well as planning staffing distributions. 

However, there are barriers to AI utilization. Many studies are designed with only data collection in mind, rather than application. So, while the data is promising, few systems have been successfully incorporated into the healthcare setting. Data accuracy is another concern. Datasets containing incorrect inputs could lead to flaws in AI decision-making. AI accuracy can be further compromised with bias, especially considering the subjectivity of some data like patient history. Finally, there are ethical concerns, particularly data sharing. Efforts must be made to de-identify data and protect patient privacy. Given that these kinks are worked through, you may see the benefits of this new technology the next time you visit the ED. 

References:

Mueller, B., Kinoshita, T., Peebles, A., Graber, M. A., & Lee, S. (2022). Artificial intelligence and machine learning in emergency medicine: A narrative review. Acute Medicine & Surgery, 9(1), e740. https://doi.org/10.1002/ams2.740