Skip to main content

Table 3 DL signature, clinical model, and nomogram predictive performance

From: A CT-based interpretable deep learning signature for predicting PD-L1 expression in bladder cancer: a two-center study

 

Training set

Validation set

 

DL signature

Clinical model

Nomogram

DL signature

Clinical model

Nomogram

AUC(95%CI)

1(0.971–1.000)

0.681(0.593–0.761)

1(0.971–1.000)

0.857(0.745–0.932)

0.528(0.398–0.655)

0.857(0.745–0.932)

ACC

1

0.677

1

0.825

0.603

0.825

SEN

1

0.610

1

0.892

0.541

0.892

SPE

1

0.780

1

0.731

0.684

0.731

P-value

< 0.001*

Reference

< 0.001*

0.001*

Reference

0.001*

  1. AUC area under the curve, ACC accuracy, CI confidence interval, SPE specificity, SEN sensitivity
  2. * In comparison with the Clinical model
Back to article page

Cancer Imaging

ISSN: 1470-7330

Contact us