Validating Virtual Reality Surgical Training Through Modeling Learning Processes with Hidden Markov Models for Workforce Development

Training is a constant source of improvement for safe and reliable workforce development. One key area of focus is medical training, where the challenge is to safely and effectively improve training methods without compromising patient health. Virtual reality (VR) presents a promising solution, particularly for surgical training. However, two research questions arise: how can we accurately quantify and measure the growth of trainees' learning processes? How can we address the heightened concerns of validity in VR training, especially when assessing how trainees are learning without the use of a physical system? The authors propose a reinforcement learning framework, utilizing Hidden Markov Models (HMM), to provide insights into the learning growth of trainees. This model incorporates five states: novice, advanced beginner, competent, proficient, and expert. The observed states are derived from a combination of performance scores and completion times during VR training. Moreover, by constructing separate models for VR-based and real-world learning data, we investigate the validity of the VR-based training program and demonstrate its effectiveness in medical training. As a result, the proposed research is expected to provide valuable insights into the VR learning process and establish its validity as a training model, while also laying the groundwork for developing optimal guidance strategies in future VR training programs.