Skip to main content

Table 4 Comparison of truth table statistics for the test sets for two models

From: Artificial neural network models for prediction of intestinal permeability of oligopeptides

Network

1 : 1 Data set

 

1 : 3 Data set

architecture

SEa

SPb

PPVc

NPVd

Acce

 

SEa

SPb

PPVc

NPVd

Acce

(7 × 20)-0-1

74

67

69

72

70

 

32

94

65

81

79

(7 × 8)-0-1

70

72

71

70

71

 

19

96

59

78

76

  1. a SE = Sensitivity : the proportion of all intestinal barrier-permeable peptides correctly predicted, SE = TP/(TP + FN) where TP is the number of intestinal barrier-permeable peptides correctly predicted and FN is the number of intestinal barrier-permeable peptides incorrectly predicted as impermeable peptides.
  2. b SP = Specificity : the proportion of intestinal barrier-impermeable peptides correctly predicted, SP = TN/(TN + FP) where TN is the number of intestinal barrier-impermeable peptides correctly predicted and FP is the number of intestinal barrier-impermeable peptides incorrectly predicted as permeable peptides.
  3. c PPV = Positive Predictive Value : the probability that a predicted permeable peptide is in fact a barrier- permeable peptide, PPV = TP/(TP + FP).
  4. d NPV = Negative Predictive Value : the probability that a predicted intestinal barrier-impermeable peptide is in fact impermeable peptide, NPV = TN/(TN + FN).
  5. e Acc = Accuracy : the percentage of all predictions that are correct, Acc = (TP + TN)/Total.