Table 11.

Permutation tests of Table 7 results (accuracy).

Cross-validation	10	20	50	100	200	400	800	1200	2000	3000	4000	6000	All (12,440)
MEAN ACROSS 100 PERMUTATIONS
2 Sample t-test	0.630	0.620	0.573	0.564	0.575	0.591	0.606	0.613	0.617	0.620	0.621	0.622	0.626
Nested CV	0.626	0.623	0.583	0.565	0.577	0.595	0.605	0.610	0.616	0.621	0.621	0.621	0.626
Recursive FE	0.592	0.593	0.609	0.615	0.618	0.620	0.625	0.625	0.625	0.624	0.626	0.626	0.626
SD OF MEAN ACROSS 100 PERMUTATIONS
2 Sample t-test	0.002	0.001	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002
Nested CV	0.002	0.003	0.002	0.002	0.002	0.002	0.002	0.002	0.001	0.002	0.002	0.002	0.002
Recursive FE	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002	0.002
Test set	10	20	50	100	200	400	800	1200	2000	3000	4000	6000	All (12,440)
MEAN ACROSS 100 PERMUTATIONS
2 Sample t-test	0.557	0.569	0.570	0.541	0.544	0.558	0.565	0.570	0.575	0.580	0.582	0.586	0.582
Nested CV	0.558	0.553	0.540	0.529	0.532	0.536	0.556	0.561	0.565	0.575	0.577	0.581	0.582
Recursive FE	0.557	0.555	0.562	0.568	0.575	0.584	0.583	0.582	0.579	0.580	0.581	0.582	0.582
SD OF MEAN ACROSS 100 PERMUTATIONS
2 Sample t-test	0.002	0.003	0.004	0.004	0.004	0.004	0.003	0.003	0.003	0.003	0.003	0.003	0.003
Nested CV	0.003	0.003	0.003	0.004	0.004	0.003	0.004	0.004	0.003	0.004	0.003	0.003	0.003
Recursive FE	0.004	0.004	0.004	0.003	0.003	0.003	0.003	0.003	0.004	0.003	0.003	0.003	0.003

Results summarizing ADHD prediction using combination of all feature types (non-imaging, anatomical, and network features extracted from SIC networks) with stratification by gender in the permutation testing framework. Entries indicate the accuracy for classifiers built using different feature selection methods (rows) and different numbers of features (columns). Top: results on leave-out folds during cross-validation. Bottom: results on separate test set based on training across all examples in the training/cross-validation set.