AI-Generated Personalized Visualization of the Safe Place in Virtual Reality vs Traditional Safe Place Imagery: Randomized Controlled Trial.

backgroundRelaxation techniques, such as the "safe place" imagery exercise, are simple and accessible strategies to cope with the negative effects of stress. While virtual reality (VR) has been applied in relaxation research, it is still unclear whether it enhances the relaxing effect of such exercises or is merely an alternative and similarly effective form of delivery.

objectiveThis study aimed to evaluate whether the use of VR enhances the relaxing effect of the safe place imagery exercise by generating an individualized virtual environment representing each participant's personal safe place with AI. Additionally, the roles of imagery vividness, presence, and satisfaction with the AI-generated safe place were examined.

methodsThis randomized controlled trial involved a single laboratory session with 4 measurement points followed by a 5-day online follow-up period involving daily self-report questionnaires. A total of 60 adults were randomly assigned to the experimental condition or the control condition. In the experimental condition, participants engaged in the safe place imagery exercise in which they imagined a place where they feel completely safe and comfortable, followed by a stress induction task. Then, participants completed a 3-minute relaxation period within an AI-generated virtual environment representing their personal safe place, which was created based on their individual descriptions. In the control condition, the same procedures were implemented; however, in the 3-minute relaxation period, participants imagined their safe place with their eyes closed. The primary outcome was self-reported relaxation, assessed with the Relaxation State Questionnaire. Secondary outcomes included imagery vividness (assessed with the Vividness of Visual Imagery Questionnaire), presence (assessed with the Igroup Presence Questionnaire), satisfaction with the AI-generated safe place, and use of the safe place in everyday life.

resultsThe 3-minute relaxation period of both conditions enhanced self-reported relaxation levels with no statistically significant differences between conditions, indicating that relaxation in VR did not outperform traditional mental imagery. Relaxation levels during the 5-day online follow-up period likewise did not differ significantly between conditions, and there were no significant differences in the frequency with which participants used their safe place during the follow-up period. Imagery vividness was positively associated with postintervention relaxation in both conditions, whereas satisfaction with the AI-generated safe place and presence in the virtual environment did not predict outcomes.

conclusionsWhile relaxation in the AI-generated safe place in VR was effective in promoting relaxation, it did not demonstrate a benefit compared to traditional mental imagery, either immediately after the intervention or during the follow-up period. These findings suggest that VR may represent an effective alternative form of delivery, but not a superior enhancement, for brief relaxation interventions. Future developments might increase the potential of AI and VR as effective tool for promoting relaxation.

trial registrationOpen Science Framework 10.17605/OSF.IO/BP53A; https://osf.io/bp53a/overview.