Student Presenter(s): Matthew Brett
Faculty Mentor: Stephen DiRusso
Department: Clinical Sciences
School/College: College of Osteopathic Medicine, Long Island

Background and Objectives: Injury of the Common Bile Duct is a severe adverse outcome of Laparoscopic Cholecystectomy. A major reported risk factor for this injury is the misidentification of the anatomical structures (cystic duct, cystic artery, common bile duct) within the Critical View of Safety (CVS). A new artificial intelligence software, Cholecystectomy AI/Surgeon's JARVIS, evaluates the CVS and has potential to aid in the identification of critical structures. This study compares human versus JARVIS evaluation of the CVS and whether JARVIS score correlates with the relative experience of the evaluators.

Methods: 25 photos of purported CVS were analyzed using JARVIS. The photos were scored by the evaluators using a published 6-point scale (6PS) for analyzing the CVS. Medical students (3), residents (2), and an attending surgeon (1) were used as the evaluators for the photos. Spearman's statistical analysis was used to determine the correlation between the subjects' 6PS and JARVIS data.

Results: There was a statistical correlation between the evaluators' 6PS and JARVIS in scoring the CVS (ranges 0.39 to 0.49). The more experienced evaluators had a stronger and significant correlation (p>0.02).

Conclusion: JARVIS has predictive validity in evaluating the Critical View of Safety during Laparoscopic Cholecystectomies. More data is needed to better establish this correlation.