Table 2 Classification matrix for prediction models used for training.

From: Integrating multi-omics and clinical features to model survival in epithelial ovarian cancer subtypes

Class

Precision

Recall

F1-score

Algorithm

AUC

0

0.81

0.72

0.77

Logistic regression

0.776

1

0.60

0.71

0.65

Logistic regression

0.776

0

0.81

0.74

0.77

SVM

0.783

1

0.61

0.70

0.65

SVM

0.783

0

0.82

0.73

0.78

Random forest

0.79

1

0.61

0.72

0.66

Random forest

0.79

0

0.78

0.80

0.79

GBC

0.81

1

0.64

0.61

0.62

GBC

0.81

  1. Class 0 = alive at last follow-up; Class 1 = deceased. SVM was selected as the optimal model based on superior recall for mortality class (0.70) and balanced F1-score (0.65), prioritizing accurate identification of high-risk patients. AUC–ROC area under the receiver operating characteristic curve.