Table 3.
Classification performance with feature selection.
| Feature Set | sens. (%) | spec. (%) | accu. (%) | AUC | (sens+spec)/2 (%) | p-value |
|---|---|---|---|---|---|---|
| FV+Demo | 78.1 ± 2.9 | 57.3 ± 2.3 | 68.6 ± 1.7 | 0.71 ± 0.01 | 67.7 ± 1.7 | NA |
| fALFF1+Demo | 75.2 ± 2.2 | 58.8 ± 2.5 | 67.7 ± 1.5 | 0.72 ± 0.02 | 67.0 ± 1.5 | 0.12 |
| fALFF2+Demo | 66.1 ± 3.3 | 48.4 ± 3.4 | 58.0 ± 2.8 | 0.64 ± 0.02 | 57.2 ± 2.8 | 3.22E-12 |
| GM+Demo | 67.0 ± 3.2 | 50.9 ± 2.4 | 59.7 ± 2.3 | 0.64 ± 0.02 | 58.9 ± 2.2 | 1.13E-12 |
| WM+Demo | 69.5 ± 3.1 | 54.9 ± 4.4 | 62.9 ± 3.0 | 0.66 ± 0.02 | 62.2 ± 3.1 | 2.05E-7 |
| CSF+Demo | 70.1 ± 1.8 | 55.7 ± 1.9 | 63.5 ± 1.4 | 0.67 ± 0.02 | 62.9 ± 1.4 | 1.25E-8 |
The best sens., spec., accu., AUC and (sens+spec)/2 are marked by bold.