JMIR Publications

ABSTRACT

Background:

Major bile duct injuries during laparoscopic cholecystectomy (LC) are a significant source of morbidity, mortality, disability, and healthcare costs. These injuries are primarily due to errors in surgical judgment and visual misperception of critical anatomy and tissue planes. In surgical education, serious games have been shown to improve intraoperative skills and decision-making.

Objective:

To facilitate learning of safe LC we developed an engaging and educational mobile game that provides coaching and feedback from an artificial intelligence (AI) model to improve intraoperative decision-making during LC.

Methods:

LapBot Safe Chole was designed to be an innovative, free, accessible and interactive simulated learning experience with real-time AI-generated feedback for deliberate practice. After downloading the game on their smartphone device, users were presented with intraoperative LC scenarios (short video clips) and tasked with identifying the ideal target zone of dissection. After submitting a response, users are provided with an accuracy score according to the pre-annotated safe and unsafe zones of dissection. Annotations come from a validated AI algorithm that provides real-time scoring and feedback overlay on the video clips and avoids the need for faculty availability. There are 5 levels in the game, with increasing difficulty based on the Parkland grading scale for cholecystitis. The game application generates a total score for each level as well as a global total score, each of which can be viewed on a leaderboard. In this study version of the game, a user is required to pass (>50% accuracy score) five consecutive questions in a row before proceeding to the next level.

Results:

Beta-testing (n = 22) results indicate improvement in game scores with each round, with attendings and senior trainees reaching top-scores earlier than junior residents per level. Candidates can be distinguished by their learning curves and learning progression which can facilitate a competency-based curriculum. There is a positive and statistically significant association between the level of the user and the accuracy of their scores (p=0.003). Specifically, the more advanced users tend to achieve higher scores on average. Qualitative feedback from our end-users was highly positive; 80% of users reported LapBot was easy to use (20% were neutral), 60% agreed it was easy to navigate (40% strongly agreed), and 100% of users reported LapBot was a very fun way to learn about safe LC.

Conclusions:

LapBot Safe Chole is a new and innovative mobile application game that integrates safe LC principles into an educational, accessible and free game using AI-generated feedback. Initial beta-testing suggests validity evidence in the assessment score and that this has a high likelihood of adoption and engagement by surgical trainees. Future directions include additional validation with a larger sample size and cross-platform iOS and Android versions of the game. Clinical Trial: N/A