Taken together, this study provides a realistic estimate of the classification performance that can be achieved in a large, ecologically valid, multi-site sample of OCD participants using data on regional brain structure. Our findings show that parcellated structural MRI data do not enable a good overall distinction between patients with OCD and HC. However, classifying subgroups of patients based on medication status enables fair identification at the individual subject level, which implies that medication use is associated with substantial distributed differences in brain anatomy. This underlines the need for longitudinal studies on the short-term and long-term effects of psychiatric medication on brain structure.
