Skip to main content

Table 9 Performance comparison among various gene sets

From: Discovery of prognostic biomarkers for predicting lung cancer metastasis using microarray and survival data

Gene set

Training Acc (%)

Training Asurv (%)

Test Acc (%)

32-gene (lung cancer related)

65.38

34.47

50.00 (13/26)

9-gene (SBS from the 32-gene set)

73.07

55.70

80.76 (21/26)

16-gene (EMT related, SVM ensemble)

93.59

74.81

76.92 (20/26)

11-gene (EMT and cancer related)

87.18

53.37

69.23 (18/26)

5-gene (EMT and lung cancer related)

78.25

52.33

76.92 (20/26)

1-gene (CCL16, maximum Acc)

74.36

43.58

57.69 (15/26)

1-gene (CDKN2AIP, maximum Asurv)

61.54

46.00

76.92 (20/26)

  1. The value of Asurv is 73.21% for real classes of samples in the training dataset.