Skip to main content

Table 5 Performance of the models on the hold-out test

From: Ultrasound-based radiomics machine learning models for diagnosing cervical lymph node metastasis in patients with non-small cell lung cancer: a multicentre study

Models

AUC (95% CI)

Accuracy (95% CI)

Semantic-LR

0.809 (0.772–0.973)

0.912 (0.798–0.947)

Radiomics-LR

0.840 (0.750–0.954)

0.947 (0.825–0.983)

Combineda-LR

0.833 (0.725–0.957)

0.965 (0.877–0.983)

Radiomics-RF

0.877 (0.861–0.970)

0.930 (0.814–0.965)

Combined-RF

0.901 (0.886–0.968)

0.947 (0.842–0.983)

  1. All statistics were validated in the test set of one institution
  2. AUC the area under the curve, CI confidence interval, LR logistic regression, RF random forest
  3. asemantic-radiomics combined
Back to article page

BMC Cancer

ISSN: 1471-2407

Contact us