Closing the Feedback Loop: The Feed Protocol for AI-Driven Curricular Reform in Surgical Residency.

OBJECTIVE: The American Board of Surgery In-Training Examination (ABSITE) assesses surgical resident knowledge, but manual analysis of program-wide data is time-consuming. This study introduces the Faculty Educational Evaluation and Deficiency Analysis (FEED) Protocol to evaluate whether artificial intelligence (AI) can rapidly identify program-wide knowledge gaps within an individual residency program and generate actionable remediation plans. DESIGN: A retrospective quality improvement proof-of-concept study utilizing a standardized master prompt-engineering sequence with a large language model (LLM). SETTING: A single-site, academic general surgery residency program. PARTICIPANTS: De-identified ABSITE "Topic Areas for Incorrect Answers" reports from 16 unique residents during the 2025 and 2026 examination cycles (total n = 32 reports). RESULTS: The AI extracted and categorized hundreds of data points into 8 surgical domains in under 60 seconds. The protocol identified a shift from moderate gaps in 2025 to critical cluster failures in 2026, including an 80% failure rate in "Nutritional Requirements." The AI successfully generated a "Faculty Action Roadmap" with specific, data-driven learning objectives for the operating room, didactics, and morbidity and mortality (M&M) conferences. CONCLUSIONS: The FEED Protocol provides a rapid, reproducible method for closing the feedback loop in surgical education. Utilizing AI for ABSITE analysis allows programs to transition from static curricula to precision education, addressing dynamic knowledge gaps in real-time.