Table 4.

Comparison of the best classification performance of the classifier for each pair, and whether that performance is significantly better than random.

Pair	Model	AUC	ACC (%)	SPEC (%)	SENS (%)	p-Value (AUC > Random)
NC vs. sd-aMCI	K = 8, γ = 16, C = 1	0.52	50	44	58	>0.05
NC vs. md-aMCI	K = 7, γ = 8, C = 100	0.66	61	60	62	<0.05
sd-aMCI vs. md-aMCI	K = 7, γ = 4, C = 0.01	0.54	53	53	53	>0.05