Exploring the Potential of Personalized Virtual Reality Environments for Sleep Support

Abstract

Sleep difficulties negatively impact health and well-being. Virtual reality (VR) has shown potential to support pre-sleep relaxation, yet most interventions rely on predefined environments. With advances in generative artificial intelligence (AI), it is now possible to personalize VR environments, potentially improving engagement, relaxation, and sleep. We developed an AI-supported approach that enables users to generate personalized VR environments based on desired pre-sleep emotions. We conducted a formative evaluation with 15 participants over seven nights by collecting sleep data, self-reports, and qualitative data via interviews. Results show that both predefined and personalized VR experiences positively impacted sleep quality. While no significant differences in sleep metrics were found between the predefined and personalized conditions, participants preferred the personalized approach. They cited greater emotional alignment, engagement, and agency. The results suggest that personalization primarily enhances the lived experience of pre-sleep routines, with limited direct impact on physiological sleep metrics.