Skip to main content

Table 3 Comparison of Receiver Operating Characteristic (ROC) values for stone-free and one-session success according to three machine learning methods

From: Machine learning prediction of stone-free success in patients with urinary stone after treatment of shock wave lithotripsy

 

Stone-free

One-session success

Random forest (RF)

 Sensitivity

0.74

0.81

 Specificity

0.92

0.75

 AUC

0.85

0.78

 CI (95%)

(0.75–0.94)

(0.67–0.86)

 PPV

0.82

0.79

Extreme gradient boosting trees (XGBoost)

 Sensitivity

0.75

0.80

 Specificity

0.93

0.75

 AUC

0.84

0.77

 CI (95%)

(0.74–0.93)

(0.68–0.87)

 PPV

0.78

0.79

Light Gradient Boosting Method (LightGBM)

 Sensitivity

0.78

0.79

 Specificity

0.92

0.74

 AUC

0.85

0.77

 CI (95%)

(0.73–0.93)

(0.67–0.87)

 PPV

0.81

0.78

  1. AUC area under ROC curve, CI confidence interval, PPV positive predictive value