Table 4 Performance evaluation of three models with RFE Applied.

From: Predicting low density lipoprotein cholesterol target attainment using machine learning in patients with coronary artery disease receiving moderate-dose statin therapy

Dataset

Metrics

Value [95% CI] 

  

RFE-applied XGBoost

RFE-applied

RF

RFE-applied

LR

Training

(n = 6,581)

ACC

0.646

[0.642–0.651]

0.600

[0.595–0.606]

0.590

[0.582–0.602]

SEN

0.642

[0.627–0.656]

0.874

[0.865–0.883]

0.910

[0.900–0.922]

SPE

0.649

[0.633–0.666]

0.347

[0.337–0.357]

0.293

[0.281–0.306]

PPV

0.630

[0.622–0.637]

0.554

[0.550–0.558]

0.544

[0.539–0.552]

NPV

0.662

[0.657–0.666]

0.748

[0.736–0.762]

0.779

[0.758–0.809]

AUROC

0.706

[0.697–0.716]

0.690

[0.681–0.699]

0.700

[0.688–0.714]

Test

(n = 2,821)

ACC

0.661

[0.643–0.678]

0.585

[0.567–0.603]

0.575

[0.557–0.594]

SEN

0.659

[0.635–0.684]

0.856

[0.836–0.875]

0.898

[0.883–0.914]

SPE

0.662

[0.639–0.687]

0.333

[0.307–0.357]

0.276

[0.253–0.298]

PPV

0.644

[0.620–0.670]

0.544

[0.523–0.565]

0.535

[0.515–0.557]

NPV

0.677

[0.652–0.701]

0.714

[0.677–0.748]

0.745

[0.706–0.781]

AUROC

0.709

[0.689–0.728]

0.691

[0.671–0.710]

0.684

[0.665–0.703]

  1. Abbreviations: ACC, accuracy; AUROC, area under the receiver operating characteristic; LR, logistic regression; ML, machine learning; NPV, negative predictive value; PPV, positive predictive value; RF, random forest; RFE, the recursive feature elimination; SEN, sensitivity; SPE, specificity; XGBoost, extreme gradient boosting;.
Back to article page