Trust in AI Clinical Decision Support Systems: A Scoping Review of Measurement and Behavior

Trust in artificial intelligence clinical decision support systems is critical to appropriate reliance, yet its measurement and behavioral implications remain unclear. We conducted a scoping review of quantitative studies published between January 2010 and June 2026, searching PubMed, PsycINFO, Web of Science, Scopus, and Embase. Thirty-three studies met inclusion criteria. Fourteen (42.4%) explicitly defined trust, and 16 (48.5%) assessed anticipated or hypothetical trust without direct system interaction. Twenty-seven studies (81.8%) examined factors associated with trust, emphasizing human–AI interaction and system characteristics over organizational and social contexts. Trust patterns varied across healthcare professionals, patients, and general populations, although direct comparisons were rare. Twenty-seven studies assessed separate AI-related behavioral outcomes, but only 11 formally tested trust–behavior relationships. Reported trust did not consistently correspond to preferences or decisions when AI and clinician recommendations conflicted. More comparable measures and observed behavioral outcomes are needed to evaluate appropriately calibrated trust in clinical AI.