Artificial Intelligence-Based Simulation to Improve Code Status Discussions Among Internal Medicine Residents: A Pilot Randomized Trial.

BACKGROUND: Code status discussions (CSDs) are essential in clinical practice yet training modalities are often resource-intensive and not widely available. AIM: To evaluate whether artificial intelligence (AI)-driven simulation can enhance CSD training. SETTING: A public safety-net hospital affiliated with an academic medical center. PARTICIPANTS: Postgraduate year (PGY)-2 and PGY-3 internal medicine residents. PROGRAM DESCRIPTION: Residents attended a lecture on effective CSDs. Post-randomization, intervention residents completed 1 h of supervised CSD practice with ChatMD, a novel AI chatbot, with faculty debriefing, followed by 1 h of independent practice. Controls continued with usual clinical training. Eight weeks (SD ± 2 weeks) later, enrolled residents completed a blinded standardized-patient encounter graded on a modified, validated checklist. PROGRAM EVALUATION: Thirty-one of 41 eligible residents enrolled; 25 completed the study. Intervention residents scored higher than controls on overall CSD performance (72.5% vs. 66.0%, p = 0.35), particularly on CSD-specific skills (71.4% vs. 57.6%, p = 0.15), but differences were not statistically significant. Qualitative analysis found that intervention residents valued ChatMD for deliberate practice and increased confidence in CSDs, while noting areas to improve emotional realism and feedback. CONCLUSIONS: Findings from this pilot feasibility study show the promise of AI-based simulation training for enhancing CSD skills, and support its continued development.