NIH researchers presents risks and benefits of integrating AI into medical decision-making process

Researchers at the National Institutes of Health (NIH) found that an artificial intelligence (AI) model accurately answered medical quiz questions designed to test healthcare professionals' diagnostic skills using clinical images and brief text summaries. However, physician reviewers noted that the AI often made mistakes in describing images and explaining its reasoning. These findings, published in *npj Digital Medicine*, highlight the potential and limitations of AI in clinical settings.

The study, conducted by NIH’s National Library of Medicine and Weill Cornell Medicine, involved the AI model and physicians answering questions from the New England Journal of Medicine’s Image Challenge. While the AI frequently matched or outperformed physicians in closed-book scenarios, it struggled with image interpretation and justification.

The research underscores the need for further evaluation of AI before integrating it into clinical practice, emphasizing that understanding AI's limitations is essential for harnessing its potential to improve medical decision-making.