Using clustering of biometric data in evaluating virtual reality experiences

Virtual reality (VR) has the potential to offer an excellent opportunity for truly immersive experiences.
However, it can sometimes be challenging to discern changes in emotional state during immersive scenarios. It might be helpful to consider the use of a stress device to detect changes during the VR experience. This study investigates the potential effectiveness of an entirely in-house developed VR roller coaster simulation, consisting of a moving chair and a visor, providing a 3D scenario with adjustable speed and level of realism. During the ride, the participants kindly agreed to use a wearable stress detection device developed in-house, which is designed to collect biometric data using heartbeat and galvanic skin (i.e. sweat level) sensors. A comparative analysis of different clustering techniques (K-means, Agglomerative, Mean Shift and Gaussian Mixture Model) has been conducted using the biometric data with the aim of identifying the various levels of stress experienced by participants during the ride. At the conclusion of the VR experience, the participants were respectfully invited to complete a brief questionnaire to share their perceptions. These data were then cross-referenced with the stress levels obtained by the clustering to check for potential correspondences, crucial to assess the effectiveness of the VR experience. This will provide insight into whether VR experiences can have an impact on emotional states and consider the potential for VR to provide a comparable experience to reality.