Virtual reality & artificial intelligence: Future of nontechnical skills training

Editor's Note

Researchers led by Florida A&M University and Florida State University in Tallahassee propose multi-learner virtual reality (VR) environments assisted by artificial intelligence (AI) as effective alternatives to facility-based training for nontechnical skills.

As failure to communicate is cited as the reason for 22% of preventable complications in the OR, the improvement of nontechnical skills “reduces technical errors and increases patient safety.” VR and AI-assisted training can alleviate these errors, and facilitate greater adoption of simulated training by guiding instructors and reducing time burdens.

Benefits. Facility-based training is often underutilized as it is more costly, and requires more staff and a dedicated space that must be maintained or turned over for different scenarios. Alternatively, a VR environment combined with an AI agent who can observe team members’ decisions and respond in real time allows learners greater accessibility and more variation in virtual patients than manikins. AI technology will additionally aid instructors in annotation and debriefing.

Effectiveness. Studies investigating effectiveness of virtual ORs observed a substantial increase in team members’ knowledge of processes and procedures. In a fire emergency scenario, only 4.4% responded correctly in their first attempt with no AI guidance. Trial 3 observed an increase to 11% and on trial 4, 31%, with assistance of AI guidance.

Barriers. Technical barriers exist in developing the simulations, as poor software performance could be associated with quality of experience and learner’s performance. There are also costs associated with acquiring VR-based training equipment. Learning with a VR headset may be uncomfortable for some users. Development of the AI requires significant data collection to properly learn and adapt to users, as limited training data could result in AI failure.

The authors conclude, “we anticipate an additional 4 to 5 years for the AI to mature to provide effective feedback to teams for individual surgical procedures.”