5 - (SYM 34) Prescriptive Predictors of Mindfulness Ecological Momentary Intervention for Social Anxiety Disorder: Machine Learning Analysis of Randomized Controlled Trial Data

Background: Shame and stigma often prevent individuals with social anxiety disorder (SAD) from seeking costly psychotherapies, highlighting the need for brief, low-cost treatments. Creating prescriptive outcome prediction models is crucial for identifying which clients might benefit most from scalable treatments. Classical regression methods may not capture complex associations, so precision medicine approaches were used to examine predictors of a 14-day self-guided mindfulness ecological momentary intervention (MEMI) via a self-monitoring app.

Method: The present study involved 191 participants in Singapore with probable SAD, randomly assigned to MEMI (n=96) or self-monitoring (n=95). Participants completed self-reports at baseline, post-treatment, and one-month follow-up. Machine learning (ML) models with 17 predictors were evaluated to identify optimization to MEMI over self-monitoring, defined as higher SAD remission probability. Models included random forest and support vector machines with 10-fold nested cross-validation.

Results: ML models outperformed logistic regression. Multivariable models using the top ten predictors achieved good performance, with area under the receiver operating characteristic curve (AU-ROC) values ranging from .71 to .72. Significant predictors included higher trait mindfulness, lower SAD severity, university education, no psychotropic medication use, higher generalized anxiety severity, clinician-diagnosed depression/anxiety, and Chinese ethnicity. Emotion dysregulation and current psychotherapy were inconsistent predictors.

Conclusion: The AU-ROC values indicated moderately meaningful effect sizes in identifying prescriptive predictors. Focusing on client strengths, weaknesses, and socio-demographics may enhance prediction of responses to scalable treatments. A 'prescriptive predictor calculator' could help clinicians allocate resources effectively. Clients with high remission probability may benefit from MEMI as a waitlist strategy before intensive therapy, aiding in creating actionable treatment selection tools for SAD care in routine healthcare settings.