3 - (SYM 24) Deriving Data Patterns from Ecological Momentary Assessment to Inform Prescription of Dialectical Behavior Therapy Skills

Growing rates of mental health concerns of the general public combined with additional stressors associated with the COVID-19 pandemic (Nochaiwong et al., 2021; Vahratian et al., 2021) suggests a dire need for increasing access to mental health resources. Brief, low-intensity, self-help interventions that can be accessed by more individuals without requiring attendance of mental health appointments may help extend the reach of care and prevent mental health concerns from developing or exacerbating (Kazdin, 2017). Nevertheless, low utilization and high attrition rates are common for self-help interventions (e.g., Fleming et al., 2018) posing barriers for individuals to maximally benefit from these interventions. One potential solution is personalizing the delivery of technology-based self-help interventions, which better gauges individuals thereby enhancing mental health outcomes.

In the current study, we explore whether pre-intervention information of individuals can predict their engagement in a video-based intervention involving a range of behavioral coping skills in order to facilitate personalized skills recommendations. The data were collected from 99 college students (Mean age = 20.69; 80.81% female; 44.44% White, 40.40% Asian, 5.05% Black, and 9.15% other races) who received brief animated videos based on behavioral skills from Dialectical Behavior Therapy (DBT) for 14 consecutive days while completing ecological momentary assessments (EMA) before (week 1-2) during (week 3-4), and after (week 5-6) the intervention (Rizvi et al., 2022).

Analyses are underway to examine whether unsupervised classifications of subgroups based on pre-intervention EMA data of positive and negative affect, self-efficacy of emotion management, and bearability of feelings can predict whether individuals watched or rewatch each video and whether they tended to watch or rewatch more acceptance- or change-oriented skills videos. In order to derive data-driven subgroups, we are working on conducting personalized analysis of each individual’s dynamic associations among the measured variables nested within subgroup (and group) structures via Group Iterative Multiple Model Estimation (GIMME). We hope that the findings will inform ways to more precisely prescribe certain skills to interested individuals and facilitate optimal sequencing of skills within the intervention.