We investigated which brain regions (features) contributed most to OCD vs. HC classifications for site-stratified CV only, using the feature importance values from RFC and permutation testing, and as we were interested in brain regions rather than the influence of the covariates, we focused on classifications using brain data only following multiple linear regression of confounding factors. No features were selected consistently (survived false discovery rate (FDR) correction in >50% CV-folds) for the main analyses (OCD patients vs. HC classification) in either pediatric, adult or combined samples. However, for medicated vs. unmedicated OCD classification in combined samples, 24 significant and consistently selected features were found. In addition, 12 features were found for early vs. late AO patients classifications in combined samples. A complete overview of these findings (including features importance for medication classification in adult samples) can be found in supplementary Tables S10–11.
