Table 4 Comparative analysis of data augmentation methods for TAAD rupture prediction.

From: Acute type A aortic dissection rupture factors prediction using a hybrid ensemble model with feature selection and data augmentation

Method

Synthetic samples

JSD

EMD

AUC

F1-score

No augmentation

0

–

–

0.852

0.821

SMOTE

900

0.162

0.842

0.927

0.861

ADASYN

900

0.178

0.901

0.919

0.849

SMOTE-ENN

850

0.155

0.813

0.931

0.868

VAE

900

0.121

0.632

0.942

0.883

CTAB-GAN+

900

0.087

0.351

0.961

0.902

CTGAN

900

0.093

0.292

0.978

0.919

Back to article page