Skip to main content

Table 3 Prediction performance of machine learning models

From: Gadoxetic acid-enhanced MRI for identifying cholangiocyte phenotype hepatocellular carcinoma by interpretable machine learning: individual application of SHAP

 

AUC (95% CI)

Accuracy (95% CI)

Sensitivity (95% CI)

Specificity (95% CI)

RF

 Training cohort

0.808(0.806–0.810)

0.774(0.771–0.777)

0.700(0.693–0.707)

0.812(0.805–0.819)

 Internal validation cohort

0.709(0.699–0.719)

0.739(0.730–0.748)

0.698(0.681–0.715)

0.760(0.742–0.778)

 External validation cohort

0.829(0.827–0.831)

0.804(0.799–0.808)

0.712(0.700-0.723)

0.856(0.844–0.868)

 Prospective validation

0.813(0.810–0.815)

0.831(0.828–0.834)

0.762(0.757–0.767)

0.860(0.855–0.865)

KNN

 Training cohort

0.834(0.830–0.837)

0.798(0.794–0.801)

0.719(0.711–0.727)

0.839(0.832–0.845)

 Internal validation cohort

0.709(0.699–0.720)

0.772(0.763–0.780)

0.654(0.638–0.670)

0.839(0.824–0.853)

 External validation cohort

0.859(0.856–0.862)

0.846(0.843–0.848)

0.744(0.734–0.753)

0.904(0.897–0.911)

 Prospective validation

0.754(0.750–0.758)

0.820(0.817–0.822)

0.653(0.644–0.662)

0.891(0.885–0.898)

SVM

 Training cohort

0.720(0.694–0.746)

0.814(0.807–0.821)

0.657(0.628–0.686)

0.895(0.889–0.902)

 Internal validation cohort

0.609(0.595–0.622)

0.751(0.744–0.758)

0.522(0.503–0.542)

0.870(0.857–0.883)

 External validation cohort

0.728(0.702–0.753)

0.797(0.788–0.805)

0.681(0.651–0.711)

0.863(0.851–0.875)

 Prospective validation

0.690(0.667–0.713)

0.752(0.739–0.766)

0.713(0.690–0.735)

0.769(0.746–0.793)

XGBoost

 Training cohort

0.835(0.826–0.843)

0.788(0.779–0.798)

0.774(0.762–0.786)

0.796(0.783–0.809)

 Internal validation cohort

0.830(0.817–0.842)

0.816(0.808–0.824)

0.799(0.780–0.818)

0.827(0.815–0.839)

 External validation cohort

0.816(0.808–0.824)

0.778(0.770–0.786)

0.775(0.760–0.790)

0.780(0.763–0.796)

 Prospective validation

0.776(0.771–0.782)

0.797(0.790–0.803)

0.693(0.681–0.705)

0.841(0.830–0.853)

LR

 Training cohort

0.790(0.788–0.792)

0.776(0.772–0.780)

0.689(0.678–0.699)

0.821(0.811–0.832)

 Internal validation cohort

0.764(0.755–0.773)

0.788(0.780–0.795)

0.715(0.700–0.730)

0.826(0.812–0.839)

 External validation cohort

0.795(0.794–0.795)

0.845(0.844–0.846)

0.597(0.594–0.599)

0.986(0.983–0.989)

 Prospective validation

0.802(0.801–0.803)

0.844(0.841–0.846)

0.684(0.679–0.689)

0.912(0.907–0.917)

  1. Notes: AUC, the area under curve, RF, random forest, KNN, K-nearest neighbor, SVM, support vector machine, XGBoost, eXtreme Gradient Boosting, LR, logistic regression. CI, confidence intervals
Back to article page

BMC Cancer

ISSN: 1471-2407

Contact us