Table 4 Predictive performance of the recurrence clinical model, radiomics model, and radiomics–clinical combined model.

From: Preoperative radiomics models for predicting risks of distinct recurrence patterns in pancreatic ductal adenocarcinoma based on contrast enhanced CT

Model type

Cohort

Signature

AUC

95% CI

Sensitivity

Specificity

Recurrence

Train

Clinical

0.723

0.6453–0.8009

0.694

0.685

Recurrence

Train

Radiomics

0.798

0.7290–0.8674

0.722

0.783

Recurrence

Train

Combined

0.801

0.7319–0.8703

0.653

0.859

Recurrence

Test

Clinical

0.711

0.5902–0.8317

0.682

0.719

Recurrence

Test

Radiomics

0.757

0.6497–0.8638

0.523

0.906

Recurrence

Test

Combined

0.760

0.6494–0.8698

0.591

0.812

Local

Train

Clinical

0.785

0.6908–0.8782

0.667

0.769

Local

Train

Radiomics

0.742

0.6390–0.8452

0.385

0.962

Local

Train

Combined

0.808

0.7179–0.8975

0.821

0.635

Local

Test

Clinical

0.748

0.5960–0.8992

0.615

0.875

Local

Test

Radiomics

0.743

0.5944–0.8912

0.500

1.000

Local

Test

Combined

0.760

0.6135–0.9057

0.577

0.937

Distant

Train

Clinical

0.784

0.6808–0.8881

0.545

0.909

Distant

Train

Radiomics

0.871

0.7875–0.9550

0.939

0.705

Distant

Train

Combined

0.896

0.8252–0.9668

0.758

0.886

Distant

Test

Clinical

0.756

0.5828–0.9298

0.677

0.778

Distant

Test

Radiomics

0.720

0.4944–0.9464

0.677

0.778

Distant

Test

Combined

0.810

0.6632–0.9569

0.516

1.000

Back to article page