Skip to main content

Table 6 Comparison of the performance of different classifiers for healthy and unhealthy discrimination

From: Efficient quantitative assessment of facial paralysis using iris segmentation and active contour-based key points detection with hybrid classifier

Classifier

Sensitivity(%)

Specificity(%)

Harmonic mean(%)

RLR

96.4

96.7

96.5

SVM

95.5

81.7

88.1

DT

95.0

96.7

95.8

NB

92.9

100.0

96.3

hDT_RLR

98.6

96.7

97.6

hDT_SVM

98.6

8.3

87.3

hDT_NB

95.0

6.7

95.8

hLR_SVM

95.6

78.3

86.1

hLR_DT

95.0

96.7

95.8

hLR_NB

95.0

96.7

95.8

hSVM_LR

95.6

96.7

96.1

hSVM_DT

92.9

96.7

94.7

hSVM_NB

92.9

100.0

96.3

hNB_LR

92.9

96.7

94.7

hNB_DT

92.9

96.7

94.7

hNB_SVM

98.6

78.3

87.3

  1. Classification methods used in the experiments. Regularized Logistic Regression (RLR), Support Vector Machine (SVM), Decision Tree (DT), Naive Bayes (NB), combined DT+RLR (hDT_RLR), DT+SVM (hDT_SVM), DT+NB (hDT_NB), LR+SVM (hLR_SVM), LR+DT (hLR_DT), LR+NB (hLR_NB), SVM+LR (hSVM_LR), SVM+DT (hSVM_DT), SVM+NB (hSVM_NB), NB+LR (hNB_LR), NB+DT (hNB_DT), NB+SVM (hNB_SVM)