At the Harvard AI Symposium this fall, Dr. Nan Du and an interdisciplinary research team from Boston Children’s Hospital unveiled MedTutor AI, an innovative artificial intelligence platform designed to improve how clinicians learn to diagnose, manage, and follow patients with complex medical conditions including celiac disease.
This innovative project, funded by the Celiac Disease Foundation, addresses one of the most pressing challenges in medical training: limited exposure to uncommon conditions and insufficient opportunities to practice diagnostic reasoning in a realistic, low-risk setting.
Celiac disease affects an estimated 1 in 100 people, yet most medical trainees receive minimal instruction on the condition. Many providers will go years before treating their first case, leading to delayed diagnosis, misdiagnosis, or missed opportunities to educate patients on management of lifelong disease. MedTutor AI aims to close that gap.
MedTutor AI changes the calculus by using generative AI to simulate realistic patient encounters, provide structured feedback, and guide learners through diagnosis, counseling, and follow-up. By recreating the types of clinical encounters pediatric GI fellows might experience in clinic, MedTutor AI gives trainees the chance to learn, make decisions, and receive detailed feedback without needing live standardized patient sessions.
So, how does it work? MedTutor AI guides the learner through a virtual patient journey in four customizable stages:
- Initial Visit: Learners ask questions, gather history, and begin building a differential diagnosis.
- Diagnosis Discussion: The system responds to learner decisions and prompts deeper clinical reasoning.
- Follow-Up Visit: Trainees manage the patient’s ongoing care and complications.
- Personalized Feedback: The platform provides rubric-based scoring, identifies strengths, and highlights missed reasoning or communication opportunities.
The program is currently being tested with real pediatric GI trainees, and early feedback has been overwhelmingly positive.
The MedTutor prototype was developed through a collaborative process that combined technical expertise from Boston Children’s Innovation Digital Health Accelerator (IDHA), celiac-focused patient scenarios created by pediatric GI specialists, iterative clinical and technical review, and usability testing with pediatric GI fellows. Ongoing enhancements including features like speech-to-text demonstrate how medical, educational, and AI innovation can come together to shape a powerful new frontier in clinical training.
Looking ahead, the team plans to expand MedTutor AI with additional celiac-specific case scenarios, compare its effectiveness to traditional learning methods, evaluate long-term knowledge retention, and adapt the tool for other pediatric conditions. Dr. Du shared, “We are incredibly excited by the potential of MedTutor AI, and look forward to expanding availability of the application nationally to help improve diagnostic recognition of celiac disease”. While still in development, MedTutor shows strong potential to deliver scalable, high-quality clinical simulation that can enhance training, especially for conditions that trainees may rarely encounter in practice.
This work reflects the Celiac Disease Foundation’s commitment to closing the diagnostic gap. By supporting MedTutor AI’s early development, the Foundation is helping bring celiac disease education into the next era, one where scalable accessible simulation gives every medical trainee meaningful exposure to celiac care.