Skip to main content

Table 4 Comparison between predictive indices of NB-BLCA models and ordinary NB in real-world data of GC patients

From: A Bayesian latent class extension of naive Bayesian classifier and its application to the classification of gastric cancer patients

 

Model

Index

NB-BLCA

(EM algorithm)

NB-BLCA

(Gibbs sampling)

NB classifier

Accuracy (95% CI)

77.22 (73.64-80.53)

87.77 (84.87-90.29)

74.71 (71.02-78.15)

No information rate (NIR)

63.32

50.92

53.43

P-Value [Accuracy > NIR]

<0.001

<0.001

<0.001

Kappa

0.53

0.76

0.49

Mcnemar's Test P-Value

<0.001

<001

0.74

Sensitivity

81.28

82.89

71.94

Specificity

74.87

92.83

77.12

Pos Pred Value

65.2

92.31

73.26

Neg Pred Value

87.35

83.95

75.93

Balanced Accuracy

78.07

87.86

74.53