Skip to main content

Table 4 Classification performance of final classifiers trained on the whole dataset.

From: Supervised prediction of drug-induced nephrotoxicity based on interleukin-6 and -8 expression levels

 

Classifier

HPTC1

HPTC2

HPTC3

Mean

AUC

TC

0.94

0.81

0.82

0.85

 

SVM

0.98

0.89

0.96

0.94

 

RF

1.00

1.00

1.00

1.00

Balanced accuracy (%)

TC

90.2

80.7

71.30

80.7

 

SVM

94.3

82.7

89.3

88.7

 

RF

98.2

99.8

100.0

99.3

Sensitivity (%)

TC

90.9

77.3

63.6

77.3

 

SVM

89.6

69.6

85.9

81.7

 

RF

96.4

99.6

100.0

98.6

Specificity (%)

TC

89.5

84.2

79.0

84.2

 

SVM

99.0

95.8

92.6

95.8

 

RF

100.0

100.0

100.0

100.0

  1. (TC=threshold-based classifier from Li et al., SVM=support vector machine, RF=random forest classifier, AUC=area under ROC curve. The values were obtained by training and test each classifier using all compounds.)