The Digital Standardized Patient: An Artificial Intelligence Coach for Cultural Dexterity in Surgical Care

BACKGROUND: Cultural dexterity, defined as the ability to effectively respond to diverse patient backgrounds, is crucial for equitable surgical care. Standardized patient (SP) encounters are proven tools for cultivating these skills but are resource-intensive and limited in scope. Large language models (LLMs) offer a scalable solution. This study evaluates the feasibility and perceived effectiveness of an LLM-based SP platform (SP-LLM) for cultural dexterity training among general surgery residents. STUDY DESIGN: In this 2-week pilot cohort study, 16 general surgery residents at a single academic center interacted with a novel SP-LLM platform simulating 2 scenarios from the Provider Awareness and Cultural Dexterity Toolkit for Surgeons curriculum. Residents engaged with the SP-LLM via both text and voice to navigate complex, culturally sensitive patient encounters. Postinteraction surveys assessed perceptions of realism, emotional depth, cultural relevance, and educational utility using Likert scales and qualitative responses. RESULTS: Residents found the SP-LLM to be clinically accurate (mean score 4.27) and its portrayals realistic (4.00). The platform supported culturally sensitive communication (mean score 3.87) and helped residents address patient emotions (3.87). Text-based interactions were valued for encouraging reflection (mean score 4.00), whereas voice interactions received more mixed feedback (3.13). Participants reported strong utility for discussing pain management (mean score 4.40) and some effectiveness in addressing medical mistrust (3.40). Qualitative feedback emphasized a need for more nuanced patient histories and improved voice realism. CONCLUSIONS: This pilot study demonstrates the promise of SP-LLMs as scalable tools for advancing cultural dexterity training in surgical education and beyond. Residents found the platform clinically relevant and effective for practicing culturally sensitive communication. With continued development, SP-LLMs have the potential to broaden access to high-quality, scenario-based training across medical specialties and institutions.