Table 5.
Classification results using independent and dependent feature sets. Classifications based on cortical thickness and age. Acc=accuracy, Sen=sensitivity, Spe=specificity, AUC= area under the ROC curve, CI=95% confidence interval.
| Classification | Independent feature sets
|
Dependent feature sets
|
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Acc (%) | Sen (%) | Spe (%) | AUC [CI] (%) | McNemar’s test | Acc (%) | Sen (%) | Spe (%) | AUC [CI] (%) | McNemar’s test | |
| pMCI36 vs. sMCI | 72.4 | 48.3 | 77.6 | 63.7 [51.5–75.9] | p<0.001 | 79.1 | 72.4 | 80.6 | 85.4 [78.4–92.4] | p<0.001 |
| pMCI24 vs. sMCI | 67.2 | 55.7 | 72.4 | 70.7 [62.7–78.7] | p=0.001 | 75.4 | 70.5 | 77.6 | 82.0 [75.7–88.3] | p<0.001 |
| pMCI12 vs. sMCI | 70.6 | 72.7 | 68.7 | 76.3 [70.4–82.1] | p<0.001 | 74.1 | 76.6 | 71.6 | 82.0 [76.9–87.0] | p<0.001 |
| pMCI6 vs. sMCI | 74.6 | 72.1 | 76.9 | 81.1 [75.8–86.5] | p<0.001 | 78.9 | 77.1 | 80.6 | 86.0 [81.4–90.7] | p<0.001 |
| AD vs. CN | 85.5 | 80.4 | 89.8 | 92.0 [89.3–94.7] | p<0.001 | 87.4 | 82.5 | 91.6 | 93.1 [90.7–95.6] | p<0.001 |
| pMCIa vs. sMCIa | 67.8 | 64.6 | 70.0 | 68.2 [62.7–73.6] | p<0.001 | 68.3 | 67.7 | 68.7 | 74.7 [69.8–79.7] | p<0.001 |
a
Cohort defined as in (Wolz et al., 2011).