Toward robust and sustainable digital mental health: Real-world validation of an AI-enabled vibro-acoustic smartphone application for stress management

Abstract Mental disorders are highly prevalent worldwide, creating demand for scalable digital tools to support psychiatric care. While many smartphone applications for stress management exist, most rely solely on self-reported outcomes and lack rigorous validation. We conducted a three-phase real-world study of an AI-enabled vibro-acoustic smartphone application. Phase 1 assessed subjective outcomes (depression, anxiety, stress, insomnia, mood, fatigue) over one month; Phase 2 measured salivary cortisol as an objective biomarker; and Phase 3 gathered user feedback (n = 1,487) after one week. Participants in Phase 1 were randomized in a 2:1 ratio (application: control) to maximize power in the intervention group, and salivary cortisol was collected once upon awakening to minimize diurnal variation, though this approach has acknowledged limitations. Results indicated improvements in depression, anxiety, insomnia, and mood in the application group, particularly among participants with high fatigue, as well as a significant reduction in cortisol levels. User feedback further suggested positive effects on relaxation, sleep, and willingness to continue use. These findings suggest that vibro-acoustic smartphone applications may contribute to stress reduction and psychological well-being in real-world settings. However, conclusions should be drawn cautiously given the modest sample sizes, reliance on a single cortisol measurement, and short-term follow-up. Larger and longer-term studies with multi-day biomarker protocols are needed to establish robustness and generalizability, but the present results highlight the potential of rigorously evaluated digital tools as complementary approaches in mental health care.