An AI Enabled Training Tool for Complex Surgical Suturing Skills

Boston Ai Llc, Medfield MA

Investigators

Abstract

Abstract Gravitate AI proposes to develop a novel Artificial Intelligence (AI)-led training tool for the assessment and development of suturing skills, designed to provide objective, standardized, and immediate feedback for surgical trainees in a convenient and self-sufficient manner outside of the operating room. Suturing is a vital surgical technique that requires skilled precision, dexterity, and efficiency and constitutes a major component of technical proficiency for both open and minimally invasive surgeries. Mastering the art of suturing technique is pivotal across numerous surgical specialties involving complex suturing of vital structures often at difficult angles and various depths in order to produce the best outcomes for patients; however, surgical trainees are limited in both time and opportunities to gain technical practice and receive feedback on their performance. Simulation models are one avenue by which to address the increased demands on modern surgical education by shifting more training to controlled settings. However, despite increasing efforts to develop suitable tools, there remains a significant gap in simulator technologies for complex suturing techniques, especially at depth, and even fewer existing technologies that can provide rapid, quantitative feedback. There is a great need for such tools to empower training surgeons in the operating room and produce the best outcomes for patients. The proposed work will build on Gravitate AI’s previously developed prototype, which consists of proprietary hardware and algorithms that analyze both static images of sutured material and dynamic recorded video footage of surgeons in action. The algorithms detect key features of suturing technique such as speed and economy of motion, while also evaluating suturing accuracy and uniformity via image segmentation based on predefined criteria determined by experienced surgeons. If successful, the project will result in an improved and validated tool which innovates significantly over traditional assessments of technical skill, mostly consisting of subjective feedback from attending surgeons. Our AI-powered tool offers a more standardized, yet personalized approach that enables comprehensive analysis and feedback based on nuanced and evidence-based metrics. The project has 2 primary Aims: 1) develop an improved AI algorithm for processing both live-feed and recorded videos of practicing surgical trainees, and 2) demonstrate proof-of-concept for consistent practice with the tool to increase key performance metrics—namely suturing uniformity, time to completion, and motion economy—in unexperienced resident surgical trainees over a 3-month period. This work will thereby support the advancement of a transformative AI training tool that will greatly help trainees effectively practice and perfect critical suturing skills, as well as equip supervisors with a reliable method to objectively track their trainees’ skill level and progress, which will result in superior outcomes for patients undergoing surgical operations. Results will directly inform a Phase II follow-on effort, involving longitudinal efficacy studies of surgical trainees using ex vivo models, as well as R&D expansions to other surgical contexts.