Who completes digital mindfulness prompts? An exploratory prediction study
Machine learning examined participation in a brief anxiety study. The outcome was completed prompts, so the findings do not establish symptom improvement.
Based on the published abstract. The full paper may contain additional methods, results and limitations.
The 30-second takeaway
The models predicted greater engagement with digital mindfulness than with self-monitoring and identified baseline features associated with that difference. Engagement meant completing prompts, rather than reduced anxiety. These exploratory findings may guide questions about participation, but they do not establish a validated way to match individuals to treatment.
The authors identify a small sample, a brief study period and reliance on a single engagement measure. The abstract does not report validation in a separate participant sample.
What researchers found
Model-predicted difference in prompt-completion engagement
Mindfulness versus self-monitoring over 2 weeks: a predicted advantage of 1.447 standard deviations on the rescaled (log-transformed) prompt-count measure. This expresses the gap relative to variation in engagement, not extra prompts completed or anxiety relief.
How prediction was tested
Researchers combined clinical, demographic and cognitive characteristics to predict participation. They used cross-validation, which tests models on held-out portions of the study data, to limit overfitting; this does not establish performance elsewhere.
Predicting a difference between interventions
Lower anxiety severity and higher attentional control predicted greater engagement with mindfulness relative to self-monitoring. These features predicted the treatment contrast; they do not necessarily indicate higher participation within the mindfulness group itself.
The original publication
Who engages? Machine learning insights into digital mindfulness-based intervention for generalized anxiety disorder.
Zainal NH, Newman MG
J Affect Disord · 2026
- PubMed ID
- 41422949
- 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.