Automated Capture of Intraoperative Adverse Events Using Artificial Intelligence

This systematic literature review was conducted to assess the current state of artificial intelligence (AI) implementation in the automatic detection of intraoperative adverse events (iAEs) during surgery. The researchers followed PRISMA-DTA standards and included articles from all surgical specialties that reported real-time, automatic identification of iAEs.

The review identified 2,982 studies, of which 13 were included for data extraction. These studies focused on detecting various iAEs, including bleeding, vessel injury, perfusion deficiencies, thermal damage, and EMG abnormalities. A meta-analysis of the AI algorithms showed promising results, with high sensitivity and specificity across the included iAEs (detection OR 14.74, CI 4.7-46.2). However, the researchers noted significant heterogeneity in the reporting of outcome statistics and article bias risk. They concluded that there is a need for standardization in iAE definitions, detection methods, and reporting to improve surgical care. The study also highlighted the versatile nature of AI technology in this field and suggested further investigation into its applications across various urologic procedures to assess the generalizability of the findings.