Artificial Intelligence (AI)-Initiated Clinical Decision Support in Hospitalist Practice

This study will evaluate the feasibility of using artificial intelligence (AI)-generated clinical decision support during routine inpatient hospital medicine care. Participating hospitalist physicians may receive AI-generated recommendations for eligible hospitalized patients under their care. Eligible patient cases will be randomly assigned to one of three AI prompting strategies that differ in the clinical information provided to the AI system. After reviewing each recommendation, participating physicians will complete a brief survey assessing the clinical usefulness of the recommendation, how they interpreted it, and whether it was or will be incorporated into patient management. The study will assess whether AI-generated recommendations can be successfully generated and delivered during routine clinical care and whether physicians complete the study surveys and adopt the recommendations. It will also provide preliminary information about physicians' perceptions of AI-generated recommendations across the three prompting strategies. The findings will be used to inform the design of a future larger randomized trial of AI-initiated clinical decision support.