Table 1 Performance of ML models and DL models based on WES data with non-coding mutations.

From: Deep learning model accurately classifies metastatic tumors from primary tumors based on mutational signatures

Model

F1 score (%)

Recall (%)

AUC (%)

AUPRC (%)

SVM

45.5 ± 0.02

30 ± 0.01

64.2 ± 0.01

79 ± 0.03

RF

83.3 ± 0.02

77.9 ± 0.04

84.8 ± 0.01

89 ± 0.09

XGBoost

83.5 ± 0.05

80 ± 0.04

84.7 ± 0.06

88.4 ± 0.04

Logistic

81.1 ± 0.01

76.9 ± 0.05

82.6 ± 0.02

86.9 ± 0.02

DiaDeL

69.3 ± 2.27

64.5 ± 8.38

77.1 ± 0.27

84.4 ± 0.12

MetaWise

86.1 ± 0.01

83.8 ± 0.03

86.8 ± 0.01

90 ± 0.01

  1. SVM support vector machine; RF random forest.
  2. The highest value for each metric is shown in bold.
Back to article page