Table 4. Performance comparison of different datasets.
| Performance | Method | Dataset | |||
| E (52) | M (305) | META-22+M (327) | E+M (357) | ||
| Accuracy (%) | wHLFS | 68.99(5.53) | 65.17(9.38) | 64.85(5.97) | 69.68(8.77) |
| wHLFS+SVM | 65.22(6.40) | 60.41(10.35) | 61.17(10.27) | 72.10(9.17) | |
| wHLFS+RF | 71.00(5.10) | 65.57(10.03) | 66.30(7.96) | 74.76(7.68) | |
| Specificity (%) | wHLFS | 78.97(11.54) | 75.88(13.26) | 73.38(12.85) | 77.13(15.80) |
| wHLFS+SVM | 71.51(13.91) | 65.22(13.21) | 64.12(11.21) | 76.51(14.60) | |
| wHLFS+RF | 77.61(12.26) | 75.88(9.88) | 75.96(11.74) | 81.43(12.99) | |
| Sensitivity (%) | wHLFS | 56.76(9.10) | 52.20(14.65) | 54.51(12.03) | 60.55(10.82) |
| wHLFS+SVM | 57.53(8.40) | 54.51(14.77) | 57.75(16.25) | 66.76(7.63) | |
| wHLFS+RF | 62.86(9.23) | 53.08(15.23) | 54.73(11.94) | 66.65(11.57) | |
MCI converter/non-converter classification comparison with different datasets in terms of accuracy, sensitivity and specificity. Methods applied here include the combinations of wHLFS and different classification methods. The different feature datasets are META (E), MRI (M), and META without baseline cognitive scores (META-22). Parameters are selected by five-fold cross validation on the training dataset. The number in the parenthesis indicates the number of features in the specific dataset. The bolded and underlined entry denotes the best performance for that particular method. The standard deviations are shown in the parentheses along with the accuracy.