Skip to main content

Table 2 Comparison with state-of-the-art methods on ROSMAP and BRCA datasets

From: MOCAT: multi-omics integration with auxiliary classifiers enhanced autoencoder

Method

ROSMAP (2 Categories)

BRCA (5 Categories)

ACC(%)

F1(%)

AUC(%)

ACC(%)

F1\(\_\)w(%)

F1\(\_\)m(%)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

KNN

65.7 (61.2-70.2)

67.1 (61.6-72.6)

70.9 (65.3-76.5)

74.2 (71.2-77.2)

73.0 (70.1-75.9)

68.2 (65.1-71.3)

SVM

77.0 (74.0-80.0)

77.8 (75.8-79.8)

77.0 (73.8-80.2)

72.9 (70.7-75.1)

70.2 (68.3-72.1)

64.0 (61.9-66.1)

Lasso

69.4 (64.8-74.0)

73.0 (68.9-77.1)

77.0 (72.7-81.3)

73.2 (71.7-74.7)

69.8 (67.9-71.7)

64.2 (61.0-67.4)

RF

72.6 (69.0-76.2)

73.4 (70.8-76.0)

81.1 (78.7-83.5)

75.4 (74.3-76.5)

73.3 (72.1-74.5)

64.9 (63.3-66.5)

XGBoost

76.0 (70.3-81.7)

77.2 (71.6-82.8)

83.7 (80.0-87.4)

78.1 (77.1-79.1)

76.4 (75.2-77.6)

70.1 (68.0-72.2)

NN

75.5 (72.9-78.1)

76.4 (73.8-79.0)

82.7 (79.6-85.8)

75.4 (71.9-78.9)

74.0 (69.8-78.2)

66.8 (61.0-72.6)

GRridge

76.0 (71.8-80.2)

76.9 (73.3-80.5)

84.1 (81.2-87.0)

74.5 (72.5-76.5)

72.6 (70.2-75.0)

65.6 (62.5-68.7)

BPLSDA

74.2 (71.2-77.2)

75.5 (72.6-78.4)

83.0 (79.9-86.1)

64.2 (63.1-65.3)

53.4 (51.7-55.1)

36.9 (34.8-39.0)

BSPLSDA

75.3 (71.2-79.4)

76.4 (72.1-80.7)

83.8 (81.2-86.4)

63.9 (62.9-64.9)

52.2 (50.2-54.2)

35.1 (32.4-37.8)

CF

78.4 (77.0-79.8)

78.8 (78.2-79.4)

88.0 (87.4-88.6)

81.5 (80.5-82.5)

81.5 (80.4-82.6)

77.1 (76.0-78.2)

GMU

77.6 (74.5-80.7)

78.4 (76.4-80.4)

86.9 (84.9-88.9)

80.0 (75.2-84.8)

79.8 (72.3-86.7)

74.6 (67.4-81.8)

Mogonet

81.5 (78.6-84.4)

82.1 (79.4-84.8)

87.4 (85.9-88.9)

82.9 (80.7-85.1)

82.5 (80.5-84.5)

77.4 (75.3-79.5)

Dynamics

84.2 (83.6-84.8)

84.6 (84.3-84.9)

91.2 (90.9-91.5)

87.7 (87.6-87.8)

88.0 (87.8-88.2)

84.5 (84.3-84.7)

MOCAT(Ours)

87.6 \(^*\) (86.7-88.5)

87.5 \(^*\) (86.8-88.2)

92.3 \(^*\) (91.2-93.4)

88.5 \(^*\) (88.1-88.9)

88.9 \(^*\) (88.5-89.3)

86.2 \(^*\) (85.3-87.1)

  1. Means and 95% confidence intervals (95% CIs) are presented, and the best results are in bold. The 95% CI is calculated using the t-distribution, with degrees of freedom set at \(n-1\), where n is the number of experiments conducted.
  2. Compared to the suboptimal model, the superior model is denoted by \(^*\) to indicate a statistically significant improvement (\(P<0.05\)) when using the two-sample t-test