An algorithm was developed to automate the grading of surgery-related adverse events using the Clavien-Dindo classification, showing substantial agreement with manual annotations. In a study of 1,379 colorectal procedures, the algorithm achieved a kappa of 0.77 with surgeons and 0.74 with nurses, improving to 0.89 after correcting annotations. The findings suggest promising potential for real-time implementation in healthcare settings, enhancing the efficiency of tracking postoperative complications through electronic health record data.
Journal Article by Båverud Olsson L, Parkan D (…) Buchli C et 3 al. in Ann Surg
Copyright © 2025 Wolters Kluwer Health, Inc. All rights reserved.