AI-Powered Prevention and Early Intervention for Stress, Depression, and Anxiety: A Randomized Clinical Trial
Abstract
Randomized evidence for purpose-built generative AI mental health support remains limited. In this preregistered trial, 1,137 unscreened university students were assigned to 6 weeks of a generative AI-powered mobile intervention (“Flourish”) or care as usual. Across the full sample, Flourish produced more favorable trajectories for perceived stress (β = −0.03, p = 0.010), depressive symptoms (β = −0.03, p = 0.006), and anxiety symptoms (β = −0.04, p = 0.001). Among participants with lower baseline symptoms, Flourish buffered deterioration; among those with elevated symptoms, improvements were larger with Flourish for depression (d = 1.03 vs 0.56) and anxiety (d = 0.91 vs 0.60), with meaningful improvement in 52.3% vs 31.7% and 50.6% vs 38.1%, respectively. Reduced perceived stress emerged as a potential pathway to symptom improvement. To our knowledge, this is the first trial supporting purpose-built, safety-first generative AI for mental health prevention and early intervention in an unscreened population.
Related
- Clinical Decision Support for Fall Prevention in Primary Care : The eSTEPS Cluster Randomized Clinical Trial
- Associations between behavioral intervention dose and outcomes in children and adolescents with obesity: meta-regression of randomized clinical trials
- A Community-Based Behavioral Intervention and Body Mass Index in School-Aged Children : A Cluster Randomized Clinical Trial
- A Primary Care E-Health Family Intervention for Risk Behaviors in Hispanic Youths : A Randomized Clinical Trial
- Chess-based cognitive remediation training as an add-on intervention treatment for alcohol use disorder: A quasi-randomized clinical trial
BibTeX
@misc{cachia2026aipowered,
title = {{AI-Powered Prevention and Early Intervention for Stress, Depression, and Anxiety: A Randomized Clinical Trial}},
author = {Julie Youko Anne Cachia and Xuan Zhao and Wenxi Pu and Chi Liao and merel keijsers and Julian De Freitas and Rebecca Fortgang},
howpublished = {PsyArXiv},
year = {2026},
month = oct,
doi = {10.31234/osf.io/nwz9q_v8},
url = {https://osf.io/preprints/psyarxiv/nwz9q_v8/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - AI-Powered Prevention and Early Intervention for Stress, Depression, and Anxiety: A Randomized Clinical Trial
AU - Julie Youko Anne Cachia
AU - Xuan Zhao
AU - Wenxi Pu
AU - Chi Liao
AU - merel keijsers
AU - Julian De Freitas
AU - Rebecca Fortgang
PY - 2026
DA - 2026/10/15
PB - PsyArXiv
DO - 10.31234/osf.io/nwz9q_v8
UR - https://osf.io/preprints/psyarxiv/nwz9q_v8/
AB - Randomized evidence for purpose-built generative AI mental health support remains limited. In this preregistered trial, 1,137 unscreened university students were assigned to 6 weeks of a generative AI-powered mobile intervention (“Flourish”) or care as usual. Across the full sample, Flourish produced more favorable trajectories for perceived stress (β = −0.03, p = 0.010), depressive symptoms (β = −0.03, p = 0.006), and anxiety symptoms (β = −0.04, p = 0.001). Among participants with lower baseline symptoms, Flourish buffered deterioration; among those with elevated symptoms, improvements were larger with Flourish for depression (d = 1.03 vs 0.56) and anxiety (d = 0.91 vs 0.60), with meaningful improvement in 52.3% vs 31.7% and 50.6% vs 38.1%, respectively. Reduced perceived stress emerged as a potential pathway to symptom improvement. To our knowledge, this is the first trial supporting purpose-built, safety-first generative AI for mental health prevention and early intervention in an unscreened population.
ER -