Participants in the Study

Participants in a recent study conducted at an Internet psychiatric clinic in Stockholm have shown promising results after receiving 12 weeks of Internet-based cognitive-behavioral therapy (ICBT) for major depressive disorder, panic disorder, or social anxiety disorder. The cohort included 6695 patients, with 46% receiving treatment for depression, 26% for panic disorder, and 28% for social anxiety disorder.

The study, which received ethical approval from the Swedish ethical review authority, utilized self-help materials based on established CBT techniques for each condition. Therapists at the clinic, who are licensed psychologists with CBT training, engaged in asynchronous written conversations with patients and provided weekly self-assessments of primary symptoms.

Treatment outcomes were measured using standardized assessments for each disorder, with a focus on predicting treatment success based on symptom reduction. The study aimed to evaluate the predictive performance and clinical usefulness of machine learning models in guiding treatment decisions.

Data used in the study included self-rated assessments, patient activity logs during treatment, and information extracted from patients’ homework reports. Variable selection, missing data management, and algorithm selection were key methodological choices in designing the prediction models.

The study’s findings, which involved building and analyzing 1680 predictive models, highlighted the importance of early prediction in Adaptive Treatment Strategies. The evaluation of nine algorithms, including lasso regression, random forest, and linear support vector regression, demonstrated the potential for machine learning to enhance treatment outcomes in psychiatric care.

For more information on the study’s data, materials, and code, readers can refer to the supplementary information provided with the article.