AI feedback improved online therapy participation, without a clear depression-score benefit
A randomized trial found stronger engagement with automated feedback, while its main analysis found no significant difference in average depression scores between groups.
Based on the published abstract. The full paper may contain additional methods, results and limitations.
The 30-second takeaway
Adding AI feedback helped sustain participation in self-guided online therapy, but the main analysis did not find a statistically significant difference in average depression scores among the study groups. A favorable result in a separate efficacy subset should not be confused with a demonstrated benefit on the trial's primary outcome.
The abstract does not explain how the efficacy analysis set was selected. Its positive follow-up result needs to be distinguished from the null primary intention-to-treat analysis.
What researchers found
Mean depressive symptom scores
PHQ-9 scores did not differ significantly among AI-supported therapy, therapy alone and waitlist at week 7 or month 3 in the intention-to-treat analysis.
Participation and symptoms differ
Participants receiving AI feedback had higher odds of weekly participation. The program added empathic and advisory responses to the same video lessons and cognitive exercises used in the therapy-only group.
A separate analysis needs context
An additional analysis of the efficacy subset found fewer participants above a depression-score threshold with AI-supported therapy than with waitlist at follow-up. That result does not establish that AI-supported therapy outperformed therapy alone.
The original publication
Effect of AI-Based Natural Language Feedback on Engagement and Clinical Outcomes in Fully Self-Guided Internet-Based Cognitive Behavioral Therapy for Depression: 3-Arm Randomized Controlled Trial.
So M, Sekizawa Y, Hashimoto S et al.
J Med Internet Res · 2026
- PubMed ID
- 41490574
- Record checked
AI-assisted research and writing. This explains one selected publication; it is not a complete review of everything known. Our approach.
One more question, understood.
Keep track of the research you’ve explored.