Machine learning can be combined with virtual or augmented reality to train surgery residents in sensitive spinal procedures.
Further, the researchers who demonstrated the technique showed their algorithms capable of finely distinguishing between junior- and senior-level surgeons by assessing their dexterity during the procedures.
The study was conducted at McGill University in Montreal and published online ahead of print in the Journal of Bone and Joint Surgery.
Vincent Bissonnette, MD, and colleagues enrolled 22 experienced surgeons and 19 trainees to perform a virtual hemilaminectomy, a minimally invasive procedure used to alleviate spinal compression.
The researchers recorded, at 20-millisecond intervals, the position, angle and force of simulated surgical instruments as the participants used these tools to remove simulated tissue.
From these metrics the team drew data with which to train five AI algorithms.
The best of these, a support vector iteration, differentiated the junior-level physicians from their senior-level counterparts with 97.6% accuracy.
This result suggests strong potential for a combined AI-virtual reality curriculum “to provide safer training and objective assessment of surgical skills, which could lead to improved patient care,” the authors conclude.
The study is available in full for free via the Ovid database.