Enhancing VR User Modelling with Pupillometry: An Eye-Tracking-Driven Machine Learning Approach

This study investigated user experience during artwork exploration in immersive virtual reality using eye-tracking and pupillometry data. The central component of the research is the presentation of Bitscope, an innovative application developed using UNITY technology. Preliminary findings suggest that the order of exposure to stationary and virtual reality (VR) environments may influence later visual attention and engagement. Large-scale data collected during VR sessions were used to develop machine-learning regression models for pupil-diameter variables. The models achieved high predictive performance, with accuracy rates of approximately 97–99%. These results indicate that pupillometry can support the modelling of user behaviour in immersive environments; however, validation using independent participant samples is needed before broader clinical applications can be considered. The integration of VR with eye tracking provides a promising framework for studying visual behaviour and may support adaptive educational tools, automated gaze-pattern assessment, and future biomedical decision-support systems. This study contributes to applications in digital art, museum experiences, visual art therapy, and e-health/neurorehabilitation, while laying the foundation for future statistical and machine learning analyses that link eye tracking with users’ emotional states.