Table 6.
p -Values for testing true differences in AUC for classifiers with independent feature sets including age. Values found by unpaired, two-sided non-parametric test (Delong et al., 1988). Values below the significance threshold of 5% are in bold.
| AUC | pMCI36 vs. sMCI
|
pMCI24 vs. sMCI
|
pMCI12 vs. sMCI
|
pMCI6 vs. sMCI
|
pMCIa vs. sMCIa
|
|
|---|---|---|---|---|---|---|
| 63.7 | 70.7 | 76.3 | 81.1 | 68.2 | ||
| pMCI36 vs. sMCI | 63.7 | – | – | – | – | 0.512 |
| pMCI24 vs. sMCI | 70.7 | 0.344 | – | – | – | 0.605 |
| pMCI12 vs. sMCI | 76.3 | 0.070 | 0.272 | – | – | 0.048 |
| pMCI6 vs. sMCI | 81.1 | 0.011 | 0.034 | 0.226 | – | 0.001 |
| AD vs. CN | 92.0 | <0.001 | <0.001 | <0.001 | <0.001 | <0.001 |
a
Cohort defined as in (Wolz et al., 2011).