using point-biserial correlations between dichotomous and continuous variables, phi correlation for dichotomous variables and Pearson correlation for continuous variables. Only those clinical features that were significantly correlated (Bonferroni-corrected) were investigated further by rerunning previously described classifications, but now using samples further split according to their correlated features (e.g., HC vs. unmedicated, short duration OCD patients; etc.). The use of more homogenous subsamples is expected to improve classification performance, while reducing sample size itself is expected to decrease performance. If classifications are relatively unaffected by further splitting, the correlated clinical variable did not have a large influence on the classification results.
