Use of a large language model integrated within the electronic medical record for the evaluation of surgical site infections – Northern California, 2025

Key Points

• To assess the performance of a large language model in adjudicating surgical site infections.
• Evaluated gpt-4o-mini in the context of electronic medical records.
• Measured sensitivity and specificity for surgical site infection detection.
• Compared workload reduction before and after implementation.
• Achieved 100% sensitivity in detecting surgical site infections.
• Recorded 69.4% specificity, indicating a high rate of false positives.
• Reduced manual screening workload by 66% in the evaluation process.