Multimodal Personalized Mobile Health Just-in-Time Adaptive Intervention for Occupational Stress Management: Pilot Study.

Background: As digital health solutions gain traction, there is an urgent need for effective, person-centered stress management tools for employees. Advances in wearable stress monitoring and machine learning now enable the collection of high-resolution, real-time data and the delivery of personalized interventions with respect to both timing and content. Despite this technological progress, there remains a notable paucity of mobile health (mHealth) interventions that capitalize on these capabilities to implement just-in-time adaptive interventions (JITAIs). Objective: This pilot study aimed to examine the preliminary effects of the RELAX mHealth app, which uses a JITAI framework to reduce occupational stress, with a particular focus on initial indications of stress reduction, the influence of the app's degree of personalization on these effects, and overall user experience. Methods: A 2-arm randomized trial was conducted with 46 employees across two phases of 3 weeks each. Participants used the RELAX mHealth app, which adapts intervention timing based on real-time stress data and tailors content according to users' preferences. Stress was assessed via multiple heart rate variability (HRV) parameters measured with wearable sensors, alongside subjective questionnaires and ecological momentary assessment (EMA) questionnaires. Usability and user satisfaction were also evaluated. In the second phase, a group randomization was implemented, with two groups of different degrees of in-app personalization. Results: The findings were mixed. While selected subjective stress indicators decreased from baseline to the end of the study, one physiological measure increased. Retrospective chronic stress (P=.048) and arousal related to the most recent stress event prior to EMA entry (P=.02) were reduced; however, the low frequency (LF)-to-high frequency (HF) ratio of HRV increased (P=.03). Other stress variables, such as work-related stress, valence of daily experiences, and momentary arousal, showed no change over the course of the study. No between-group differences were observed with respect to the assigned degree of personalization. Additionally, usability increased (P<.001), whereas satisfaction decreased at a trend level (P=.05). Conclusions: Although the results provide selective preliminary indications of stress reduction with app use, the variability of outcomes raises concerns about its current practical utility. The effects are also discussed with regard to the specific design of the prototype and the study setup. Technical issues and user discomfort were noted despite increasing usability over the study period, indicating areas for improvement in the developed prototype. Given the absence of a control group and the use of interventional arms only, no definitive conclusions about effects can be drawn. Nevertheless, this study highlights both the potential and the challenges of a personalized mHealth approach to occupational stress management. Further refinement of the app and more comprehensive evaluations are needed to fully understand the impact and functional usability of this JITAI-based approach in real-world settings.