Table 2.

APM model performance across 100 iterations.

	#Iter of null model	#Iter of true model	Spearman rho	One-tailed p-value
Average resting-state connectivity - DMN	1000	100	0.1792 (0.0022)	< 0.001
Average resting-state connectivity – FP	1000	100	0.1207 (0.0065)	< 0.001
Average resting-state connectivity - SAL	1000	100	0.0443 (0.0017)	0.001
Cognitive PC1 (general ability)	1000	100	0.1404 (0.0060)	< 0.001
Cognitive PC2 (executive function)	1000	100	0.0592 (0.0016)	< 0.001
Cognitive PC3 (learning/memory)	1000	100	0.0803 (0.0018)	< 0.001

Results reported here use both positive and negative features (i.e., combined model performance). Model performance is quantified as the Spearman correlation (rho) between the test-set predictions and actual values across the whole sample. Standard deviations are shown in parentheses. Bold font indicates the best performing models for each domain.