TABLE 5.
Classification performance of resample data based on RSFS algorithm and SVM classifier using the top 10 features.
| MCInc vs. MCIc | MCIc vs. AD | |||||||
| ACC(%) | SEN(%) | SPE(%) | AUC | ACC(%) | SEN(%) | SPE(%) | AUC | |
| nosampling | 84.71 | 66.67 | 94.55 | 0.888 | 89.80 | 78.95 | 96.67 | 0.854 |
| downsampling | 80.20 | 76.37 | 84.03 | 0.853 | 80.80 | 71.87 | 89.73 | 0.827 |
| upsampling1 | 91.59 | 90.00 | 93.18 | 0.953 | 91.57 | 85.65 | 97.49 | 0.947 |
| upsampling2 | 92.70 | 91.74 | 93.66 | 0.962 | 88.90 | 80.59 | 97.21 | 0.934 |
Downsampling and upsampling1 are defined as random resampling. Upsampling2 is defined as ensuring that each original sample is included, and then randomly resampling the remaining data.