Clinical interviewing is an essential skill in medical training, but often requires significant time and training. Generative artificial intelligence (AI) can be used to help accelerate this training; however, its effectiveness remains unclear. Now, a new study reports how AI-based assessment of medical interview transcripts closely matches human-based assessments. These findings suggest that AI could support more efficient and scalable training in medical education.
AI in Medical Interview Assessment
Clinical interviewing is a vital skill in medical training, but its assessment is often time-intensive and challenging to scale. Researchers from Japan investigated whether generative artificial intelligence (AI) could offer a more efficient solution by evaluating medical interview transcripts. In a cross-sectional validation study, seven participants conducted interviews with an AI-simulated patient, and the resulting transcripts were assessed by both AI models (GPT-o1 Pro and GPT-5 Pro) and five experienced clinical instructors. The study found that AI-based assessments demonstrated strong agreement with human evaluations and exhibited greater consistency across repeated assessments, while significantly reducing the time required for evaluation. This research suggests a practical 'AI-first, faculty-verified' model, where AI handles the initial assessment, allowing educators to focus on coaching and high-stakes decisions, thereby providing rapid and scalable feedback to students. However, the researchers caution that AI should be used with human oversight, as text-only scoring cannot capture nonverbal cues, tone, or cultural nuances important in real-world patient interactions.
Implications for Medical Education and Future Outlook
The findings have significant implications for medical education, potentially overcoming delays in feedback that often limit opportunities for students to improve communication skills. By providing immediate and consistent evaluations, AI could make frequent practice more accessible, especially in resource-limited settings. This could lead to more timely learning experiences for students, allowing them to engage with AI-simulated patients and receive instant feedback. While the study highlights AI's growing role, the researchers emphasize the need for cautious implementation, acknowledging the limitations of transcript-based evaluation in capturing nonverbal and cultural aspects. Future approaches combining the efficiency of AI with the critical judgment of clinicians could lead to more efficient and scalable medical training systems, addressing the rising demand for high-quality medical education.
References and Authors
The research was published on February 17, 2026, in Volume 12 of the journal JMIR Medical Education, with the DOI 10.2196/81673. The study involved multiple authors, including Dr. Hiromizu Takahashi and Professor Toshio Naito from the Department of General Medicine, Juntendo University Faculty of Medicine, Japan, along with researchers from various other institutions and departments across Japan, the United States, and The Netherlands.
About Professor Toshio Naito
Dr. Toshio Naito, MD, PhD, MBA, is a Professor in the Department of General Medicine at Juntendo University Faculty of Medicine, Tokyo, Japan. With over 30 years of clinical and academic experience, his extensive research focuses on general medicine, infectious diseases, HIV, and medical education. He has a notable publication record, including 112 original articles and 4 review articles, contributing significantly to both clinical practice and advancements in medical training.
