Table 3 Main consequences of models.

From: Establishment and validation of a ResNet-based radiomics model for predicting prognosis in cervical spinal cord injury patients

Model

Dataset

Accuracy

AUC

Sensitivity

Specificity

Clinical

Training

0.882

0.866

0.981

1

Testing

0.858

0.899

0.778

1

Validation

0.676

0.754

0.667

0.714

Radiomics

Training

0.985

1.000

0.963

0.444

Testing

0.824

0.915

0.963

0.571

Validation

0.794

0.720

0.889

0.429

Combined

Training

0.853

1.000

1

0.839

Testing

0.978

0.952

1

0.889

Validation

0.824

0.815

0.852

0.714

Back to article page