The buzz around artificial intelligence (AI) in medical education is palpable, with major conferences like the Association of American Medical Colleges (AAMC) and the International Association of Medical Science Educators (IAMSE) leaning heavily into the topic. These gatherings highlight the excitement about AI's potential to transform how future doctors are trained. However, this enthusiasm is tempered by caution, as advisory committees and working groups are being formed to address the ethical, practical, and regulatory challenges AI presents. While AI promises to enhance learning and streamline processes, it also requires thoughtful integration to ensure it complements rather than compromises medical education’s core values. Let’s explore some examples of where AI is showing some promise in medical education.
Assessing and Developing Clinical Interview Skills
At UT Southwestern Medical Center’s Simulation Center, AI has been used to transform the assessment of clinical skills. In the fall of 2023, the center deployed a generative AI-based grading system for Objective Structured Clinical Examination (OSCE) notes. This system reduced human effort by 91%, decreased grading turnaround time from weeks to days, and showed that AI achieved up to 89.7% agreement with human expert graders. This groundbreaking step demonstrates how AI can handle large-scale educational challenges while maintaining accuracy and scalability (1).
Another study, published in JMIR Medical Education (2), investigated whether AI-simulated patients could improve medical students’ interview skills. Conducted in Japan, the nonrandomized controlled trial involved fourth-year medical students engaging with large language models for simulated patient interactions. The AI intervention group scored significantly higher on medical interview assessments compared to the control group, showcasing the promise of AI-simulated training programs. However, the study also highlighted the need for complementary training to address nonverbal communication skills.
Building Clinical Reasoning Through Case-Based Learning
AI is also revolutionizing case-based learning, a cornerstone of medical training. Generative AI models can create realistic patient cases, allowing students to work through diagnoses and treatment plans in a controlled setting. A study on arXiv (3) highlights how these systems provide dynamic, interactive cases that mimic real-life scenarios.
By integrating AI into case-based learning, students can gain a deeper understanding of clinical reasoning and decision-making. These tools build critical thinking skills and help learners see the direct consequences of their decisions, enhancing their ability to handle complex clinical situations. Platforms like DDx leverage generative AI to simulate comprehensive medical scenarios, offering real-time feedback that guides students through diagnostic reasoning. These tools represent the future of case-based education by combining interactive scenarios with targeted learning objectives.
Challenges and Opportunities
Of course, integrating AI into medical education isn’t without its challenges. Concerns like data privacy, algorithmic bias, and fears of losing the “human” side of medicine need to be addressed. Schools also need to invest in the right infrastructure and train faculty to use these tools effectively.
A study published in Surgical Radiology Anatomy (4) underscores the importance of careful oversight when using AI tools for medical education. The study evaluated the reliability of ChatGPT versions 3.5 and 4.0 in teaching the anatomy of the scalenovertebral triangle. It found that responses were often inaccurate or incomplete, with none aligning perfectly with standard anatomical descriptions. This highlights the risks of students relying solely on AI without educator input, as inaccuracies could lead to misunderstandings. The authors emphasized the need for cautious planning and ongoing development to improve AI tools for medical education.
Another challenge is ensuring equal access to AI technologies. Not all students or schools have the same resources, so there’s a risk of widening educational inequalities. Institutions must focus on making these tools accessible to everyone.
The Road Ahead
AI has the potential to transform medical education in ways we’re only beginning to understand. By improving curriculum planning, enhancing clinical skills assessment, and advancing case-based learning, AI can prepare future doctors to thrive in an increasingly tech-driven healthcare system.
But it’s not just about the tech. Medical education must still emphasize empathy, ethics, and human connection. With the right balance, AI can become an invaluable partner in training the next generation of compassionate, skilled doctors.
As MedEd evolves, continued research and collaboration between educators, technologists, and healthcare professionals will be essential. Together, we can create a future where AI and humanity work side by side to deliver exceptional care.
About the Author:
Andrew Berg, MD, is the co-founder of Sketchy, an innovative visual learning platform that empowers students worldwide. He earned his Doctor of Medicine (MD) from the University of California, Irvine, and is a practicing emergency medicine physician in the Bay Area. With a passion for creative education, he has revolutionized the teaching and understanding of complex concepts through his work with Sketchy.
Sources
1. Jamieson, A. R., Holcomb, M. J., Dalton, T. O., Campbell, K. K., Vedovato, S., Shakur, A. H., Kang, S., Hein, D., Lawson, J., Danuser, G., & Scott, D. J. (2024). Rubrics to prompts: Assessing medical student post-encounter notes with AI. NEJM AI, 1(12). https://doi.org/10.1056/AIcs2400631
2. Yamamoto, A., Koda, M., Ogawa, H., Miyoshi, T., Maeda, Y., Otsuka, F., & Ino, H. (2024). Enhancing medical interview skills through AI-simulated patient interactions: Nonrandomized controlled trial. JMIR Medical Education, 10, e58753. https://doi.org/10.2196/58753
3. Chu, S. N., & Goodell, A. J. (2024). Synthetic patients: Simulating difficult conversations with multimodal generative AI for medical education. arXiv. https://doi.org/10.48550/arXiv.2405.19941
4. Singal, A., & Goyal, S. (2024). Reliability and efficiency of ChatGPT 3.5 and 4.0 as a tool for scalenovertebral triangle anatomy education. Surgical and Radiologic Anatomy, 47(1), 24. https://doi.org/10.1007/s00276-024-03513-8
