Sensor-Based Intervention Modeling: A Personalized Tool to Support Intervention in Adolescent Mood Disorders - Project Summary Mood disorders in adolescence are prevalent, disabling, and associated with future chronicity and severity of mood problems in adulthood, making this a developmental period in which effective early intervention is especially important. A critical common ingredient of many interventions for adolescent mood disorders is inter- session monitoring of behaviors and mood; such monitoring is used by patients and practitioners to support case conceptualization, identify patient-specific maladaptive behaviors, and evaluate treatment outcomes. However, adherence to inter-session monitoring is low among adolescents, undermining intervention effectiveness. To optimize and support delivery of effective interventions for adolescent mood disorders, we are in need of novel, low-burden strategies for monitoring behavior and mood. The proposed study aims to address this urgent need by validating and translating a novel smartphone-based intervention monitoring approach (Sensor-Based Intervention Monitoring, SBIM). Aims will be tested in an adolescent sample (n=33, ages 13-19 years) engaged in interventions for mood disorders through the Helen and Arthur E. Johnson Depression Center or the Child and Family Clinic at the University of Colorado Anschutz Medical Campus. Study participants will complete a baseline evaluation followed by an eight-week period of SBIM using smartphones to collect native digital sensors (e.g., GPS, accelerometer, log data of screen use, app use, call/text activity, and ambient light/sound) and ecological momentary assessment (EMA) of mood symptoms. Daily behaviors are measured by extracting behavioral features from digital sensors, and include sensor-based measures of behavioral activation, impulsivity, physical activity, social withdrawal, sleep/circadian disturbances, and goal-directedness. SBIM Reports, providing inter- session monitoring information about patient-specific behaviors, mood, and behavioral predictors of mood, are delivered to clinicians and patients biweekly. Clinicians and patients report on acceptability, appropriateness, and feasibility of SBIM, and treatment outcomes, biweekly. Aim 1 centers on rigorously validating SBIM, testing the predictive accuracy of personalized models and comparing personalized to general models. Aim 2 evaluates feasibility, appropriateness, and acceptability of SBIM for both patients and clinical providers, and Aim 3 tests convergent validity of SBIM with standard treatment outcomes. Exploratory Aim 4 explores subgroups to compare differences in SBIM outcomes between interventions (behavioral, medication management), diagnoses (unipolar, bipolar) and as a function of sex, age, or pubertal stage. Ultimately, we aim that this work will develop and validate a novel monitoring tool and explore translation to the clinic, providing a foundation for research that scales and investigates SBIM outcomes.
