- Published by Researchvlog
- 11/12/2022
- 19:50
Article published in npj Digital Medicine
Yilmaz, R., Winkler-Schwartz, A., Mirchi, N. et al. Continuous monitoring of surgical bimanual expertise using deep neural networks in virtual reality simulation. npj Digit. Med. 5, 54 (2022). https://doi.org/10.1038/s41746-022-00…
‘This work presents a comprehensive surgical technical skill monitoring system with predictive validation throughout surgical residency training, with the ability to detect errors.’
Alexander Winkler-Schwartz MD
3 VIDEOS
Recai Yilmaz MD
2 VIDEOS
Prof. Rolando Del Maestro
2 VIDEOS
Nykan Mirchi
1 VIDEOS
Aiden Reich
1 VIDEOS
In procedural-based medicine, the technical ability can be a critical determinant of patient outcomes. Psychomotor performance occurs in real-time, hence a continuous assessment is necessary to provide action-oriented feedback and error avoidance guidance. We outline a deep learning application, the Intelligent Continuous Expertise Monitoring System (ICEMS), to assess surgical bimanual performance at 0.2-s intervals. A long-short term memory network was built using neurosurgeon and student performance in 156 virtually simulated tumor resection tasks. Algorithm predictive ability was tested separately on 144 procedures by scoring the performance of neurosurgical trainees who are at different training stages. The ICEMS successfully differentiated between neurosurgeons, senior trainees, junior trainees, and students. Trainee average performance score correlated with the year of training in neurosurgery. Furthermore, coaching and risk assessment for critical metrics were demonstrated. This work presents a comprehensive technical skill monitoring system with predictive validation throughout surgical residency training, with the ability to detect errors.
The mastery of technical skills is of fundamental importance in medicine and surgery as technical errors can result in poor patient outcomes. The learning of bimanual psychomotor skills still largely follows an apprenticeship model: one defined by a trainee completing a fixed-length residency working closely with instructors. Technical skills education is transitioning from this time-focused approach to competency-based quantifiable frameworks
Surgical trainees are considered competent when they can perform specific surgical procedures safely and efficiently, encompassing knowledge, judgement, technical and social skills to solve familiar and novel situations to provide adequate patient care. The focus on “adequate” rather than “excellent” or “expert” patient care relates to challenges in outlining, assessing, quantifying, and teaching the composites of surgical expertise. To provide competency-based frameworks for complex psychomotor technical skills, advanced platforms need to be created which provide objective feedback during training along with error mitigation systems. These frameworks need to be transparent and based on quantifiable objective metrics.
A technically challenging operative procedure in surgery involves the subpial resection of brain tumors adjacent to critical cortical structures. Neurosurgical graduates are expected to be proficient in this complex bimanual skill which includes minimizing injury to adjacent normal tissues and hemorrhage from subpial vessels. Technical errors in this procedure can result in significant patient morbidity. Our group developed complex realistic virtual reality tumor resection tasks to aid learners in the mastery of this skill. Exploiting these simulations on the NeuroVR platform with haptic feedback (CAE Healthcare, Montreal, Canada) we quantified multiple components of the bimanual psychomotor skills used to expertly perform these tasks. Utilizing this data post-hoc, we developed expert performance benchmarks to which learner scores were compared and machine learning algorithms to classify participants into pre-defined expertise categories. Limitations of these applications were the inability of ongoing assessment and error detection and improving performance during the task by providing continuous feedback.
Most surgical skills learning occurs in the operating room, with the surgeon instructor continuously evaluating trainee performance and providing coaching to improve performance with a particular focus on preventing surgical errors which may cause patient injury. This assessment occurs in real-time and is relevant to the precise action being performed by the trainee and the risks associated with that action. To mimic the role of expert operative instructors, we developed an artificial intelligence (AI) deep learning application, the Intelligent Continuous Expertise Monitoring System (ICEMS). The ICEMS was developed with two objectives: 1) to make a continuous assessment of psychomotor skills to detect less-skilled performance during surgery, and 2) to provide ongoing action-oriented feedback and risk notifications.
This paper outlines the development of the ICEMS and provides predictive validation evidence that enables future studies to explore its efficacy in simulation training. To our knowledge, this application is the first continuous bimanual technical skill assessment using deep learning with the predictive validation of surgical trainee performance throughout a residency program.