TOTB-02: AUTOMATED COACHING USING EXPERT INSIGHTS FOR SURGICAL SKILLS TRAINING
Anand Malpani, PhD, Gina L. Adrales, MD, MPH; Johns Hopkins University

What problem in education is addressed by this work?:
Coaching is effective for surgical skills training, but access to individualized expert feedback is limited. Current technology for surgical skills training, e.g., simulation, is widely accessible, but it only plays a passive role in that it does not provide the learner with actionable feedback. Machine learning techniques that incorporate expert insights can transform technology into an active contributor through automation in teaching surgical skills.

Describe the intervention:
Our endpoint is to automate a virtual coach for surgical skills training. Our approach to automation involves capturing expert insights using methodologies such as the “think aloud” technique, video review, and semi-structured interviews. Using this corpus of expert insights, we build machine intelligence to automatically provide immediate and individualized feedback that evaluates performance, identifies skill deficits, and recommends deliberate practice in real-time.

Describe how this intervention could be applied at other institutions. Please specifically comment on identified barriers that could exist and how they could be overcome:
The automated virtual coach can be used directly by trainees at other institutions to receive individualized feedback during surgical skills acquisition. The current framework, which relies upon a small sample of experts, must be adapted to local surgical practice and institutional skills training curricula. This can be overcome by engaging a heterogeneous sample of expert surgeons from multiple institutions to build the machine intelligence for real-time deployment and widespread applicability.