Skip to main content

Table 3 Performance of different machine learning algorithms with reference to the DLR signature

From: CT-based deep learning radiomics nomogram for the prediction of pathological grade in bladder cancer: a multicenter study

 

Training cohort

External test cohort

AUC (95% CI)

Accuracy

AUC (95% CI)

Accuracy

LR

0.931(0.909–0.954)

0.864

0.856(0.805–0.907)

0.776

NaiveBayes

0.854(0.819–0.889)

0.787

0.884(0.836–0.932)

0.804

SVM

0.953(0.931–0.974)

0.902

0.943(0.916–0.970)

0.840

KNN

0.899(0.873–0.925)

0.819

0.799(0.740–0.859)

0.790

RandomForest

0.999(0.997-1.000)

0.983

0.778(0.718–0.837)

0.703

ExtraTrees

1.000(nan-nan)

1.000

0.828(0.775–0.881)

0.744

XGBoost

1.000(nan-nan)

0.998

0.850(0.796–0.904)

0.785

LightGBM

0.986(0.979–0.993)

0.887

0.821(0.766–0.875)

0.667

GradientBoosting

0.936(0.915–0.958)

0.832

0.817(0.756–0.877)

0.676

AdaBoost

0.872(0.840–0.904)

0.808

0.636(0.561–0.711)

0.644

MLP

0.959(0.943–0.975)

0.891

0.878(0.835–0.922)

0.781

  1. AUC, area under the curve; CI, confidence interval; LR, logistic regression; SVM, support vector machine; KNN, K nearest neighbor; ExtraTrees, extremely randomized trees; XGBoost, eXtreme Gradient Boosting; LightGBM, Light Gradient Boosting Machine; MLP, Multi-Layer perceptron
Back to article page

Cancer Imaging

ISSN: 1470-7330

Contact us