Executive Functioning as a Predictor of Digital Cognitive Behavioral Therapy Outcomes for Body Dysmorphic Disorder: A Machine Learning Approach

Prior research presents mixed findings regarding whether executive functioning (EF) issues, such as cognitive flexibility and inhibition deficits, predict cognitive behavioral therapy (CBT) outcomes in body dysmorphic disorder (BDD). Elucidating how EF domains contribute to CBT response can contribute to our understanding of why some individuals benefit more than others and may inform personalized intervention strategies. We harnessed precision medicine methods to determine key inhibition and cognitive flexibility domains hypothesized to predict treatment outcomes for BDD. We predicted treatment outcomes (response, reliable change, and remission) following 12 weeks of coach-guided smartphone-based CBT for BDD, using baseline EF features. Specifically, we evaluated five inhibition features derived from the Stroop Color Word Task, eight cognitive flexibility features from the Wisconsin Card Sorting Test (WCST), and baseline symptom severity. The final models with the top five baseline EF predictors of post-CBT outcomes yielded good multivariate predictive performance (area under the receiver operating characteristic curve = 0.723–0.768) for all examined outcomes. Better conceptual learning efficiency (WCST learning-to-learn), better cognitive efficiency (fewer WCST total trials), and poorer cognitive proficiency (lower WCST number of total correct trials) predicted greater benefits from the app-based treatment. Other inhibition and cognitive flexibility measures were inconsistent or nonsignificant predictors. Results suggest the potential value of EF domain predictors for BDD treatment outcomes and warrant replication in larger samples.