Predicting response to a smartphone-based cognitive-behavioral therapy for body dysmorphic disorder

A 2024 analysis of a randomized trial asks who is more likely to benefit from smartphone-delivered cognitive behavioral therapy for body dysmorphic disorder.

Published in the Journal of Affective Disorders (volume 355, June 2024), the study was led by Jennifer L. Greenberg with a team that includes researchers from the Center for OCD and Related Disorders at Massachusetts General Hospital, directed by Sabine Wilhelm, and from MIT: Omar Costilla-Reyes, Morgan Talbot, and Armando Solar-Lezama. Body dysmorphic disorder (BDD) is a severe, chronic condition if untreated, marked by distressing preoccupation with perceived flaws in appearance. Cognitive behavioral therapy (CBT), a structured therapy that works on unhelpful thoughts and behaviors, has been shown to work for BDD, and the abstract notes that delivering it by smartphone can reduce barriers such as the shortage of clinicians, cost, and stigma.

The journal abstract describes a secondary analysis of a randomized, waitlist-controlled trial in which participants recruited nationally used a coach-guided CBT app, some immediately and some after a waiting period. The main outcome was BDD symptom severity over time, and the secondary outcomes were treatment response, a substantial reduction in severity, and remission. Because little was known about who benefits most, the analysis asked which baseline characteristics predict these outcomes.

Starting treatment immediately rather than after the wait predicted better symptom improvement, as did gender identity. Higher baseline treatment credibility and expectancy, meaning how believable and promising participants found the treatment at the outset, predicted both response and remission; lower baseline severity predicted remission, and sexual minority status predicted response and remission.

The authors suggest that smartphone CBT may offer an advantage for historically marginalized groups, and stress hastening access, building confidence in the treatment at the start, and developing stratified care, in which treatment is matched to the patient. Knowing which baseline characteristics go with better outcomes helps match patients to the form of care that suits them, which matters for a condition where specialist clinicians are scarce. Dr. Costilla Reyes described the contribution as advancing interpretable computational methods for predicting outcomes, with the longer-term aim of giving clinicians early insight that could help stratify patients.

The analysis grew out of Dr. Costilla Reyes’s digital mental health outcome modeling project at MIT, begun in 2020. It was followed by a 2025 preprint with Talbot applying interpretable machine learning to the same clinical question, and by the 2026 study of comorbid anxiety in smartphone-delivered depression interventions.