AI may improve surgical workflow and outcomes, but supporting evidence remains limited and uneven, according to a point–counterpoint debate published in Trauma Surgery & Acute Care Open.
Proponents described emerging applications such as models to forecast supply needs, tools to optimize scheduling and automate documentation, and perioperative technologies including chatbots with geospatial mapping to guide patients to appropriate care settings. They also highlighted early-stage postoperative monitoring approaches using computer vision and time-series analysis to support earlier detection of complications, as well as educational uses such as simulation platforms and AI-assisted feedback to enhance surgical training.
A review of 25 studies cited in the analysis reported that AI use was associated with shorter operative times and recovery periods, fewer intraoperative complications, and greater surgical precision, with reported reductions of 25%, 15%, and 30%, and a 40% improvement in precision.
Opponents emphasized concerns related to trust, safety, and reliability. Many models rely on registry-based data sets that lack multimodal inputs and may show reduced performance when externally validated. Underrepresentation of women, racial and ethnic minorities, and patients in low-resource settings raises concerns about perpetuating disparities. Reports of algorithmic “hallucinations” and spurious correlations further underscore risks in clinical application.
They also cautioned that premature adoption could introduce unintended burdens, drawing parallels to early electronic health record implementations that initially promised efficiency but often added complexity and new sources of error.
A systematic review cited in the analysis reported that 45% of surgical AI models met high validation standards, whereas publicly accessible data sets were available in only 14%, raising concerns about reproducibility and transparency.
The authors noted that many studies lack external validation and that gaps remain before real-world evidence of improved outcomes is established. While certain applications—particularly in workflow optimization and postoperative monitoring—are approaching clinical readiness, most systems require further validation and robust governance prior to implementation.
Both perspectives emphasized the need for clinician involvement, rigorous validation, and inclusive data development, while differing in their assessment of whether current systems are ready for widespread clinical use.
“[T]he question is not whether AI will influence surgical care but rather how it will do so, and at what pace. The integration of AI must be deliberate, evidence-based, and patient-centered,” the authors wrote.
The authors reported no conflicts of interest.
Source: Trauma Surgery & Acute Care Open