Natural Language Processing in Health Professions Education: A Scoping Review

Significant promise exists for specific applications: The review found that NLP and AI show particular technical promise in three areas—automated assessment of student work, clinical simulation with virtual patients, and analysis of educational curricula—suggesting these are the most mature and reliable uses in health education currently.

Critical safety and fairness concerns remain unresolved: Across the 64 studies reviewed, persistent challenges include AI systems producing false information ("hallucination"), accuracy problems, algorithmic bias, and data privacy risks. These issues suggest that widespread deployment of NLP/AI in health education requires substantial additional safeguards before implementation.

Lack of standardized measurement hampers responsible deployment: The review found an "absence of standardized outcome measures" across studies, meaning institutions cannot easily compare whether these AI tools actually improve student learning or patient outcomes—a fundamental requirement for responsible adoption in healthcare education.

Public health education is dangerously under-researched: Only 4 of the 64 studies (approximately 6%) addressed public health and health promotion education, representing what the authors call "a major evidence gap." This is concerning given AI's potential impact on population health initiatives and health equity.

Risk of over-reliance without proper oversight: The review identifies "overreliance" as a documented challenge, suggesting educators and students may depend on AI tools without adequate human oversight or critical evaluation—a particular concern in healthcare contexts where errors have serious consequences.