Skip to main content
. 2025 Sep 1;16:1630863. doi: 10.3389/fimmu.2025.1630863

Table 7.

The performance of the ML models developed using SVC-L1–selected features on the validation dataset.

Model Sensitivity Specificity Accuracy AUC Kappa MCC

  • DT

  • 62.71

  • 48.49

  • 57.61

  • 0.58

  • 0.11

  • 0.11

  • RF

  • 59.32

  • 72.73

  • 64.13

  • 0.70

  • 0.29

  • 0.31

  • LR

  • 59.32

  • 66.67

  • 61.96

  • 0.71

  • 0.24

  • 0.25

  • XGB

  • 66.10

  • 69.70

  • 67.39

  • 0.69

  • 0.34

  • 0.34

  • KNN

  • 61.02

  • 69.70

  • 64.13

  • 0.71

  • 0.28

  • 0.30

  • GNB

  • 62.71

  • 66.67

  • 64.13

  • 0.68

  • 0.27

  • 0.28

  • ET

  • 69.49

  • 54.55

  • 64.13

  • 0.69

  • 0.24

  • 0.24

  • SVC

  • 62.71

  • 75.76

  • 67.39

  • 0.72

  • 0.35

  • 0.37

  • MLP

  • 62.71

  • 66.67

  • 64.13

  • 0.71

  • 0.27

  • 0.28

DT, decision tree; RF, random forest; LR, logistic regression; XGB, extreme gradient boosting; KNN, k-nearest neighbors; GNB, Gaussian Naïve Baise; ET, extra tree; SVC, support vector classifier; MLP, multilayer perceptron; AUC, area under curve; kappa: Cohen’s kappa coefficient; MCC, Mathew’s correlation coefficient.