Current Applications, Challenges, and Future Directions of Artificial Intelligence in Emergency Medicine: A Narrative Review

Artificial intelligence (AI) systems have witnessed notable advancements, revolutionizing various fields of research and medicine. Specifically, advancements of AI and the rapid growth of machine learning hold immense potential to significantly impact emergency medicine. This narrative review aimed to summarize AI applications in prehospital emergency care, emergency radiology, triage and patient classification, emergency diagnosis and interventions, pediatric emergency care, trauma care, outcome prediction, as well as the legal and ethical challenges and limitations of AI use in emergency medicine. A comprehensive literature search was conducted in Web of Science, Scopus, and Medline using a wide range of artificial intelligence and machine learning-related keywords combined with terms related to emergency medicine to identify relevant published studies. The findings show that AI-powered tools can assist clinicians in emergency departments in improving the management of prehospital emergency care, emergency radiology, triage, emergency department workflow, complex diagnoses, treatment, clinical decision-making, pediatric emergency care, trauma care, and the prediction of admissions, discharges, complications, and outcomes. However, the majority of these applications have been reported in retrospective studies, whereas randomized controlled trials (RCTs) are essential to determine the true value of AI in emergency settings. These applications can serve as effective tools in emergency departments when they are continuously supplied with high-quality real-time data and are adopted through collaboration between skilled data scientists and clinicians. Implementing these AI-assisted tools in emergency departments requires adequate infrastructure and machine learning operation systems. Since emergency medicine involves various clinical decision-making scenarios based on classifications, flowcharts, and well-structured approaches, future well-designed prospective studies are necessary to achieve the goal of replacing conventional methods with new AI and machine learning techniques.