Table 3.

Performance of logistic regression classifiers (“leave-one-out” cross-validation)

Comparison	AD versus controls**	MCI-c versus MCI-n*	MCI-c versus MCI-n (including FAQ)*
Sensitivity	94.4	84.6	89.7
Specificity	84.8	55.7	62.9
Positive predictive value	81.0	51.6	57.4
Negative predictive value	95.7	86.7	91.7
Area under ROC curve	93.6	72.8	76.5

^*McNemar's chi-square test versus random classifier, P < 0.05

^**P < 0.0001