Comparison of AI and Physician Recommendations in Virtual Care

This study is a retrospective cohort analysis that compares initial artificial intelligence (AI) recommendations with final physician decisions during AI-assisted virtual urgent care visits. Conducted at Cedars-Sinai Connect, the research involved 461 physician-managed visits where AI provided diagnosis and management recommendations based on patient data. The study aimed to evaluate the concordance between AI-generated recommendations and those made by physicians who had access to the AI suggestions. Results indicated that AI recommendations were concordant with physician decisions in 56.8% of cases. Furthermore, AI recommendations were rated as optimal more frequently than those made by physicians. The findings suggest that while AI performed better in identifying critical issues and adhering to guidelines, physicians excelled in adapting recommendations based on real-time information. The study highlights the potential role of AI in enhancing physician decision-making in virtual urgent care settings.