Summary: A first-of-its-kind clinical pilot found that a personalized, machine-learning–guided lifestyle coaching program nearly doubled remission rates for people with mild-to-moderate depression. By identifying the unique behavioral factors that predict each person’s low moods and translating those insights into tailored action plans, researchers achieved substantial symptom relief and broader cognitive and quality-of-life gains.
Using consumer smartwatches and frequent real-time self-reports, the team built individualized Mood Augmentation Plans (iMAPs) that participants followed with weekly remote coaching. After six weeks, more than half the participants no longer met clinical criteria for depression, and anxiety, cognition, and quality-of-life measures also improved—offering a scalable model for remote, personalized mental health care.
Key Facts
- Widespread need: Over 21% of U.S. adults experience depression, and many could benefit from changes in daily habits such as sleep, exercise, diet, and social contact. Standard, generic advice often fails because depression and its triggers vary greatly between individuals.
- Two-week biometric baseline: Fifty adults with mild-to-moderate depression completed a two-week monitoring phase. They wore smartwatches to capture heart rate and activity, and they logged mood and brief surveys up to four times daily about sleep, diet, activity, and social interactions.
- Machine-learning personalization: UC San Diego’s Neural Engineering and Translation Labs (NEATLabs) trained individualized machine learning models on each participant’s data to detect the lifestyle factors most strongly linked to low mood.
- Individualized Mood Augmentation Plans (iMAPs): Health coaches converted model outputs into focused, evidence-based behavioral strategies. Each participant followed a unique plan targeting their top predictive factor—for example, cognitive behavioral approaches for insomnia, activity optimization, social engagement strategies, or diet-focused interventions.
- Clinical impact: Whereas typical behavioral interventions show about a 30% remission rate, this algorithm-guided, coach-supported approach yielded a 55% remission rate at six weeks—meaning a majority no longer met depression criteria by standard screening.
- Broader benefits and durability: Participants reported a 36% reduction in anxiety symptoms, notable improvements in quality of life, and measurable gains on brief tests of memory and attention. Follow-up assessments showed these benefits were maintained for at least three months after active coaching ended.
Source: UCSD
Background: Jyoti Mishra, PhD, associate professor of psychiatry at the University of California San Diego School of Medicine, notes that lifestyle changes can help many people with mild-to-moderate depression. However, broad, one-size-fits-all recommendations are often overwhelming and ineffective because each person’s depressive triggers differ.
To address this gap, Mishra and colleagues developed a machine-learning–guided lifestyle coaching program that turns personal device data into individualized interventions. The pilot clinical trial shows that targeted, data-driven coaching delivered remotely can produce rapid and meaningful improvement.
The study, published in NPP—Digital Psychiatry and Neuroscience, enrolled 50 adults in a single-arm, open-label pilot called Personalized Mood Augmentation (PerMA). During the two-week monitoring window, participants provided ecological momentary assessments via smartphone and continuous smartwatch data on sleep, movement, and heart rate. The research team then used those signals to build N-of-1 predictive models that identified each person’s dominant lifestyle predictors of low mood.
Following model-based assessment, participants worked with trained health coaches to execute their individualized mood augmentation plans for six weeks using weekly short video sessions. Interventions were selected from established behavioral therapies and tailored to the participant’s top predictive factor.
At the end of the six-week coaching period, results among completers included:
- A large reduction in depressive symptoms: 55% of participants no longer met depression criteria on the Patient Health Questionnaire-9 (PHQ-9).
- A 36% reduction in anxiety symptoms measured by the GAD-7 scale.
- Significant self-reported improvements in quality of life.
- Better performance on brief objective tests of memory, attention, and interference processing.
Follow-up assessments showed that these improvements remained stable during a three-month period after the active coaching ended, suggesting sustained therapeutic effects.
Mishra emphasizes that personalized, data-driven coaching may be more achievable and motivating for people in depressed states than broad lifestyle directives. “When someone is in survival mode, a long list of generic changes is overwhelming. Targeting one or two individualized drivers makes change more realistic and less exhausting,” she explains.
Although this pilot was small and uncontrolled, it provides early evidence that combining digital monitoring, individualized machine-learning insights, and brief remote coaching is a promising integrated approach for treating mild-to-moderate depression at scale. The authors recommend larger, randomized controlled trials to confirm efficacy.
Additional co-authors include Jason Nan, Suzanna Purpura, Satish Jaiswal, Houtan Afshar, Vojislav Maric, James K. Manchanda, Charles T. Taylor, and Dhakshin Ramanathan. The trial received partial funding from a seed grant by the Hope for Depression Research Foundation.
Key Questions Answered:
A: Generic guidance can be overwhelming for someone in a depressed state who is focused on day-to-day survival. Depression manifests differently across individuals, so sweeping recommendations may be irrelevant or unattainable. By identifying a person’s single most predictive lifestyle driver and addressing it directly, the intervention reduces the burden of changing everything at once and improves the chance of meaningful benefit.
A: Over the two-week monitoring period, the model examines correlations between biometric signals (like activity and heart rate) and frequent mood ratings to determine which lifestyle factors most strongly predict low mood for that individual. Coaches then use those prioritized factors to select targeted, evidence-based behavioral therapies—such as CBT for insomnia when sleep emerges as the primary driver, or social engagement strategies when lack of connection is most predictive.
A: In this pilot, participants maintained large reductions in depression and anxiety and continued cognitive and quality-of-life gains during a three-month follow-up after weekly coaching concluded, indicating that effects can persist beyond the active intervention.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The underlying journal paper was reviewed in full.
- Additional context was provided by editorial staff.
About this AI and depression research news
Author: Susanne Bard
Source: UCSD
Contact: Susanne Bard – UCSD
Image: Image credited to Neuroscience News
Original Research: Open access. “Personalized machine learning guided intervention for optimizing lifestyle behaviors in depression: a pilot study” by Jason Nan, Suzanna Purpura, Satish Jaiswal, Houtan Afshar, Vojislav Maric, James K. Manchanda, Charles T. Taylor, Dhakshin Ramanathan & Jyoti Mishra. DOI: 10.1038/s44277-026-00062-3
Abstract
Personalized machine learning guided intervention for optimizing lifestyle behaviors in depression: a pilot study
There is a pressing need for personalized, data-driven interventions for depression. In this pilot study, the researchers applied N-of-1 machine learning to tailor behavioral lifestyle interventions. Fifty adults with mild-to-moderate depression enrolled in the single-arm, open-label Personalized Mood Augmentation (PerMA) pilot clinical trial (NCT05662254).
Participants completed a two-week digital monitoring phase using smartphone ecological momentary assessments (four times per day) combined with smartwatch tracking of sleep, exercise, diet, and social connection. Personalized machine learning models identified which lifestyle factors best predicted each person’s mood. Those results were translated into individualized Mood Augmentation Plans (iMAPs) that participants implemented for six weeks with weekly guidance from a health coach.
Among intervention completers (n=40), depression symptoms decreased significantly: self-rated PHQ-9 scores fell (mean change −3.5 ± 3.8, large effect size) and clinician-rated HDRS scores also improved substantially. Co-occurring anxiety reduced significantly, and quality of life showed meaningful gains. Objective cognitive measures affected by depression—selective attention, interference processing, and working memory—also improved. Ecological momentary assessment confirmed that mood improvements were specifically linked to improvements in the individualized lifestyle targets. Decision algorithms and large language models matched human coach iMAP assignment with high accuracy in this dataset.
The PerMA trial demonstrates a promising personalized lifestyle intervention approach for depression and supports further scale-up and randomized controlled trials to establish clinical efficacy. PERMA was registered on ClinicalTrials.gov under registry number NCT05662254.