Skip to main content

Table 3 Detailed performance metrics for the four models in training set

From: Clinical Data based XGBoost Algorithm for infection risk prediction of patients with decompensated cirrhosis: a 10-year (2012–2021) Multicenter Retrospective Case-control study

Models

AUROC

Sensitivity

Specificity

PPV

NPV

(95%CI)

(95%CI)

(95%CI)

(95%CI)

(95%CI)

Mutil-tree XGBoost

0.985

0.934

0.938

0.943

0.929

 

(0.982–0.987)

(0.924–0.944)

(0.928–0.949)

(0.933–0.952)

(0.918–0.940)

Simple-tree XGBoost

0.971

0.915

0.900

0.908

0.907

 

(0.967–0.975)

(0.903–0.926)

(0.887–0.913)

(0.897–0.920)

(0.894–0.919)

Logistic-11

0.869

0.712

0.878

0.864

0.738

 

(0.858–0.879)

(0.694–0.731)

(0.864–0.892)

(0.848–0.880)

(0.720–0.755)

Logistic-6

0.864

0.727

0.860

0.849

0.744

 

(0.853–0.875)

(0.709–0.746)

(0.845–0.875)

(0.834–0.865)

(0.726–0.761)

  1. AUC: area under the receiver operating characteristic curve; PPV: positive predictive value; NPV: negative predictive value; CI: Confidence Interval