Table 3 Performance of the machine-learning models for early prediction of dialysis liberation.

From: Predictive approach for liberation from acute dialysis in ICU patients using interpretable machine learning

 

Models

Sensitivity

Specificity

Brier Score

Accuracy

AUROC

2015–5019

Development and fivefold cross-validation

XGBoost

0.48 ± 0.04

0.91 ± 0.04

0.22 ± 0.05

0.78 ± 0.05

0.81 ± 0.03

RF

0.38 ± 0.02

0.95 ± 0.02

0.21 ± 0.03

0.79 ± 0.03

0.80 ± 0.02

LR

0.43 ± 0.04

0.89 ± 0.04

0.24 ± 0.05

0.76 ± 0.05

0.77 ± 0.01

2020

Temporal testing

XGBoost

0.57

0.93

0.16

0.84

0.85 (0.81–0.88)

RF

0.44

0.94

0.18

0.82

0.83 (0.80–0.87)

LR

0.44

0.91

0.21

0.79

0.82 (0.79–0.85)

  1. AUROC, area under receiver operating characteristic; LR, logistic regression; RF, random forest; XGBoost, extreme gradient boosting. Values in parentheses are 95% confidence intervals.
Back to article page