Table 1 The prediction performance of different deep learning models.

From: Development and validation of a multi-omics hemorrhagic transformation model based on hyperattenuated imaging markers following mechanical thrombectomy

Model name

AUC

95% CI

Sensitivity

Specificity

Accuracy

PPV

NPV

Precision

ResNet152

0.799

0.6538–0.9449

0.647

0.857

0.708

0.917

0.5

0.917

Inception v3

0.738

0.5889–0.8880

0.353

0.929

0.521

0.923

0.371

0.923

VGG 16

0.739

0.5888–0.8902

0.618

0.714

0.646

0.84

0.435

0.84

VGG 19

0.69

0.5207–0.8595

0.441

0.786

0.542

0.833

0.367

0.833

ResNet50

0.683

0.5159–0.8496

0.324

0.857

0.479

0.846

0.343

0.846

  1. AUC, area under curve; PPV, positive predictive value; NPV, negative predictive value; VGG, Visual Geometry Group.
Back to article page