Table 4 Generalization performance of the classifiers for the classification between HC and patients with mild cognitive impairment. In this table, the rows indicate the dataset where the classifier is trained and the columns indicate the dataset where the performance was measured. The area under the receiver operating characteristic curve is shown with the upper and lower bound of its 95% confidence interval. No case had a performance significantly different from the normative approach calculated using the confidence interval of the difference between the approach across the bootstrap scheme.
AIBL | ARWiBo | |
|---|---|---|
AIBL | – | 0.61 [0.54, 0.67] |
ARWiBo | 0.59 [0.53, 0.65] | – |