Large language model maps residents' pathology exposure gaps to personalize case-based teaching

A large language model (ChatGPT-4o) tracked radiology residents' clinical reports with >91% precision and recall, then targeted missed pathologies. Median unique pathology exposure rose in every subspecialty evaluated without significantly reducing live case reads, except postgr…

• ChatGPT-4o achieved over 91% precision and recall for identifying important pathologies from routine resident clinical reports.
• Residents encountered significantly more unique pathologies in abdominal, musculoskeletal, neuro, pediatric, and thoracic imaging after implementation.
• The supplemental teaching did not significantly reduce live case interpretation volumes, except for postgraduate year 3 abdominal imaging.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.