Physician adoption patterns of AI-driven clinical decision support systems in urinary tract infection management

In 2021, Maccabi Healthcare Services (MHS) introduced "UTI Smart-Set" (UTIS), an AI-driven decision support system (DSS) based on a machine-learning (ML) algorithm, to optimize antibiotic treatments for UTIs. UTIS reduced antibiotic mismatch-defined as pathogen resistance to prescribed empiric antibiotic based on culture-by ~ 30%, yet ~ 33% of physicians did not follow its recommendations. We aimed to characterize physicians according to UTIS implementation. We conducted a retrospective cohort study using MHS data of UTI encounters between 9/2023 and 3/2024. Analysis included 626 physicians and 15,033 encounters. We examined correlations between physicians' characteristics and implementing UTIS recommendations, accounting for patient- and encounter-level variables. Results indicted that physicians with younger patients population (odds ratio [OR], 0.952 per year, 95% CI 0.922-0.983), diagnose more UTIs (OR 1.021 per case, 95% CI 1.007-1.035), and work within group practices (OR 1.542, 95% CI 1.02-2.333), were more likely to follow UTIS recommendations. Conversely, older physicians (OR 1.034 per year, 95% CI 1.012-1.056), Arabic sector (OR 3.474, 95% CI 1.709-7.062), and a higher volume of patients (OR 1.027 per 100 patients, 95% CI 1.003-1.052) were less likely to implement UTIS recommendations. Addressing these physicians' characteristics is important to improve the integration of DSS.