Validation of ChatGPT-4o for detecting surgical site infections after colorectal surgery from electronic health records: a large-scale diagnostic accuracy study.

Key Points

• The study aims to validate the effectiveness of ChatGPT-4o in detecting surgical site infections (SSI) using electronic health records (EHR).
• Analyzed EHR narratives to identify surgical site infections post-colorectal surgery.
• Evaluated the diagnostic performance metrics including sensitivity and specificity.
• Compared performance for different types of SSI: overall, deep incisional, and organ/space SSI.
• Overall SSI detection sensitivity was moderate at 21.1%, with high specificity at 98.7%.
• Deep incisional SSI sensitivity was also low at 20%, with high specificity at 99.7%.
• Organ/space SSI detection showed improved sensitivity at 85.9% and specificity at 84.6%.