Personalized machine learning guided intervention for optimizing lifestyle behaviors in depression: a pilot study.

Area of Science:

• Digital health
• Machine learning in medicine
• Mental health interventions

Background:

• Depression requires personalized interventions.
• N-of-1 machine learning (ML) can optimize behavioral lifestyle adjustments.
• The Personalized Mood Augmentation (PerMA) trial investigated this approach.

Purpose of the Study:

• To leverage N-of-1 ML for personalized depression treatment.
• To identify individual lifestyle factors impacting mood.
• To assess the efficacy of data-driven lifestyle interventions.

Main Methods:

• 50 participants with mild-to-moderate depression used smartphone-based ecological momentary assessments (EMAs) and smartwatch tracking.
• Personalized ML models identified key lifestyle predictors (sleep, exercise, diet, social connection).
• Individualized mood augmentation plans (iMAPs) were implemented with health coach support.

Main Results:

• Significant reductions in depression symptoms (PHQ9, HDRS) and anxiety (GAD7).
• Sustained benefits up to 12-week follow-up.
• Improvements in quality of life and cognitive functions (attention, working memory).
• ML models and LLMs achieved 95% accuracy in matching iMAPs.

Conclusions:

• The PerMA trial demonstrates a viable personalized lifestyle intervention for depression.
• Data-driven N-of-1 ML can effectively target mood-influencing behaviors.
• Further RCT testing and scale-up are recommended for clinical validation.