About this project
We develop person-specific models that forecast short-term (about one hour ahead) occurrences of adolescent maladaptive symptoms (e.g., negative affect, repetitive negative thinking). Models learn from brief in-the-moment surveys (EMA) and passive sensing from phones and wearables (e.g., smartwatches). Guided by interpersonal theory, we also examine the highly individualized mechanisms — such as attachment orientation and momentary interpersonal context (warmth/dominance) — that can trigger or buffer symptoms. The current aim is high-quality prediction for each individual. In later phases, these models will inform a personalized just-in-time adaptive intervention (JITAI) – a digital support system that delivers short, targeted interventions exactly when an individual’s personal model detects they are at high risk.
Why does it matter?
A lot of adolescents struggle with their mental health, but many never get the help they need. Stigma, long waitlists, and parents’ logistics keep support out of reach—and even for those who start therapy, drop-out is common. These gaps leave many moments of distress unaddressed in daily life.
Our approach uses something nearly every teen already has—a smartphone—to detect those moments in real time. By combining EMA and passive sensing, we can not only predict when risk is about to rise, but also test why—examining mechanisms like interpersonal triggers, attachment-related sensitivities, and momentary shifts in warmth or dominance during social interactions. Understanding both timing and cause builds the scientific foundation for future just-in-time interventions that deliver the right kind of help at exactly the right moment.
How does it work?
Everyday data: 8 EMA prompts/day (up to 90 sec each) + passive sensing (that does not require any typing).
Short-term forecasting: Each participant gets a personal model that updates as data accrue and predicts symptoms in the next hour.
Context modeling: Drawing on the clinical background we gained over the years, and belief in the central role of interpersonal processes in shaping individual psychopathological states- we test whether momentary interpersonal context (e.g., warmth/dominance, attachment-relevant signals) improves predictions.