Table 4.
Comparison of performance of different classifiers distinguishing participants with MCI from those with intact cognition
| Classifier | Sensitivity | Specificity | Accuracy | AUC-ROC |
|---|---|---|---|---|
| Chance | 30.0 | 76.0 | 60.0 | 52.2 |
| Nonlinear SVM (RBF) | 53.2 | 88.2 | 76.2 | 71.2† |
| Linear SVM | 60.96 | 77.5 | 71.9 | 69.2† |
| Linear SVM + L1-norm | 72.7 | 72.4 | 72.4 | 72.5† |
| RFC | 6.51 | 72.3 | 74.7 | 68.2† |
Abbreviations: MCI, mild cognitive impairment; AUC-ROC, area under the curve of receiver operating characteristics; SVM, support vector machine; RBF, radial basis function; RFC, random forest classifier.
NOTE. †P < .05.