Ethical considerations for intraoperative implementation of artificial intelligence clinical decision support systems: a scoping review
TL;DR
An Ethics-Implementation Scorecard is proposed as an illustrative conceptual framework to structure reporting of ethical implementation domains and support future evaluation of intraoperative AI-based CDS systems.
Abstract
Intraoperative artificial intelligence (AI)-based clinical decision support (CDS) systems are increasingly proposed for surgical decision-making, yet evidence supporting their clinical implementation and ethical governance remains limited. We searched five databases for studies published between 2010 and July 2026, identifying 3,020 records. Five studies met the inclusion criteria: one completed feasibility study, and four ongoing prospective registries and randomized trials across vascular surgery, gynecologic oncology, and anesthesiology. Although most systems operated in human-in-the-loop configurations, published evidence of intraoperative performance remained limited to a single feasibility study, highlighting a substantial translational gap between technical innovation and prospective clinical validation. Across the included studies, technical validation was reported more frequently than governance, regulatory compliance, equity, and oversight considerations, while clinical evaluation remained geographically concentrated, potentially limiting generalizability. We propose an Ethics-Implementation Scorecard as an illustrative conceptual framework to structure reporting of ethical implementation domains and support future evaluation of intraoperative AI-based CDS systems.