Enhancing Immunology Education for Rheumatology Trainees Using Artificial Intelligence: A Pilot Quality Improvement Initiative

Objectives Immunology is foundational to rheumatology yet often remains a conceptual challenge for many trainees. Artificial intelligence (AI), particularly large language models (LLMs), offers the potential to provide personalized, responsive, and learner-directed education. This pilot study explored the feasibility and educational value of an AI-guided learning tool in supporting rheumatology trainees’ understanding of key immunologic principles. Methods We conducted a quality improvement project involving postgraduate year (PGY) 4 and 5 rheumatology trainees. Participants completed a baseline survey assessing self-rated confidence in immunology, pathophysiology, drug mechanisms of action (MOA), and exam preparedness on a 5-point Likert scale. Participants also completed a 10-item multiple-choice knowledge test before and after the AI session. Trainees then engaged in 2 20-minute self-directed sessions using ChatGPT with 2 structured prompts focused on disease pathophysiology and pharmacologic mechanisms. Prompts incorporated clinical relevance, analogies, visual metaphors, Socratic questioning, as well as built-in knowledge checks. After completion of the AI session, participants repeated the confidence surveys, knowledge test and provided qualitative feedback via free-text responses. Quantitative results were descriptively analyzed. Qualitative feedback was thematically coded. Results Ten trainees completed the study (6 PGY-4, 4 PGY-5). Mean test scores improved from 80% to 98%, with the most substantial gains observed among participants with lower baseline scores. Confidence ratings improved across all domains: immunology (2.2 to 3.2), pathophysiology (2.8 to 3.78), MOA (2.5 to 3.67), and exam preparedness (2.1 to 3.67). Qualitative analysis revealed strong perceived educational value, with participants citing interactivity, real-time feedback, and the ability to simplify complex content as strengths. Areas for improvement included enhanced visuals, interface design, and content accuracy verification. Conclusion This pilot study demonstrates the potential of AI-guided educational tools to improve both confidence and knowledge in immunology among rheumatology trainees. While limitations include small sample size and lack of long-term outcome data, findings support further exploration of AI as a scalable and customizable adjunct to traditional instruction. Integration of such tools may enrich specialty training and align medical education with evolving learner needs. Best Abstract By A Rheumatology Resident Award