Skip to main content
. 2021 Feb 5;15:605115. doi: 10.3389/fnins.2021.605115

TABLE 8.

10-fold cross-validation binary mean classification performance for MCI against AD using LSVM classifier using different feature selection methods.

Feature selection method Performance metrics Mean (%) Standard deviation Max (%) Min (%)
LASSO Accuracy 97.80 0.9862 98.32 91.99
Sensitivity 97.62 1.0065 97.92 91.73
Specificity 97.74 1.0720 98.5 92.07
F-measure 0.98
FSASL Accuracy 83.71 0.90 85.00 78.15
Sensitivity 90.63 1.57 92.23 84.12
Specificity 77.10 1.098 79.74 72.55
F-measure 0.838
LLCFS Accuracy 90.04 1.43 92.02 86.07
Sensitivity 90.23 1.43 91.94 86.74
Specificity 90.09 1.77 92.52 84.62
F-measure 0.903
CFS Accuracy 84.41 1.95 86.81 79.18
Sensitivity 90.37 2.14 93.26 83.73
Specificity 78.10 1.90 80.79 72.84
F-measure 0.83
SVMRFE Accuracy 82.87 0.903 83.96 78.832
Sensitivity 81.68 1.22 84.34 77.10
Specificity 84.11 0.94 85.21 80.72
F-measure 0.854