Back to 2024 Abstracts
Surgical Intelligence Can Lead to Higher Adoption of Best Practices in Minimally Invasive Surgery
*Monica Ortenzi
2, *Danit Dayan
3, *Eran Nizri
3,
Gerald M. Fried1, *Yuval Mirkin
4, *Sari Maril
4, *Dotan Asselmann
4, *Tamir Wolf
41Surgery, McGill University, Montreal, QC, Canada; 2General and Emergency Surgery,, Polytechnic University of Marche, Ancona, Italy; 3General Surgery, Tel Aviv Sourasky Medical Center; Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel; 4Theator Inc, Palo Alto, CA
Introduction: Surgical intelligence technology encompasses routine, automated capture and analysis of surgical video and connection of derived data with patient and outcome data. At its core are computer vision algorithms that recognize intraoperative actions and events. Recent work extends these capabilities to include recognition of safety measures, such as critical view of safety (CVS) in laparoscopic cholecystectomy (LC). Given its role in minimizing bile duct injury, increasing CVS adoption among surgeons is a compelling target for quality improvement intervention. We examined the use of surgical intelligence for automatically monitoring CVS adoption to enable a real-world quality initiative.
Methods: LCs conducted at two general surgery departments of one medical center between December 2022 and August 2023 were routinely captured by a surgical intelligence platform. AI-based analysis, validated by surgical annotation experts, indicated for each video whether the three major CVS components had been achieved, and duration of surgery recorded. LC case complexity estimates were derived algorithmically based on the Parkland scale. These data, both per procedure and aggregated, were accessible to surgeons and clinical leadership on the platform. In March 2023, the departments launched a quality improvement initiative aiming to reach 70% CVS adoption across all LC procedures. Outcome measures were monthly and 3-month adoption percentages for full CVS and for each component, before and after intervention. Associations with duration and complexity were assessed.
Results: The final dataset included 279 videos; overall CVS was achieved in 154 (55.2%). CVS adoption increased from 39.2% in the 3 months before intervention to 69.2% in the final 3 months (p < .001), with monthly adoption rising from 33.3% to 75.7%. The “two structures visibility� component rose from 33.3% to 81.1% (p < .001) while adoption of the other two CVS components started and remained high. Procedures with full CVS were shorter than those without (p < .001) and mean duration decreased from the 3 pre-intervention months (M = 54.1 min, SD = 38.0) to the final 3 months (M = 44.3 min, SD = 26.0; p = .039). CVS adoption was lower for higher complexity procedures but this difference was not statistically significant.
Conclusion: Introduction of a surgical intelligence platform led to a steady increase in full CVS adoption, reaching the goal within 6 months of launching the initiative. Routine, ongoing assessment also revealed that low initial adoption was related to a single CVS component and that increased adoption came with a drop in procedure duration. These results are the first to demonstrate how real-world use of surgical intelligence for routinely, automatically assessing adoption of an intraoperative safety measure can uncover new insights, modify surgeon behavior, and enable focused initiatives to implement best practices for improving the quality and uniformity of surgical care.
Back to 2024 Abstracts
