Table 3.
The table depicts each feature’s importance in descending order calculated by permutation.
0.0403 ± 0.0503 | HC |
0.0245 ± 0.0413 | RAVLT-Im |
0.0086 ± 0.0108 | AGE |
0.0058 ± 0.0141 | CFT |
0.0000 ± 0.0129 | GENDER |
− 0.0014 ± 0.0211 | APOE |
− 0.0014 ± 0.0058 | GDS |
− 0.0029 ± 0.0503 | RAVLT-delay |
− 0.0058 ± 0.0168 | TMTA |
− 0.0101 ± 0.0147 | TMTB |
− 0.0129 ± 0.0058 | ANART |
− 0.0158 ± 0.0279 | RAVLT-recog |
The leftmost column in each row depict average effect on model accuracy by random shuffling ± how the accuracy varied from one reshuffling to the next. The two most important features are hippocampal volume and RAVLT immediate, followed by age, category fluency and gender.