Skip to main content

Table 4 Pooled analysis of other evaluation metrics (accuracy, sensitivity, specificity) for readmission prediction

From: Evaluation of machine learning methods for prediction of heart failure mortality and readmission: meta-analysis

Model

Accuracy

Sensitivity

Specificity

N

Pooled AUC

Confidence interval

N

Pooled AUC

Confidence interval

N

Pooled AUC

Confidence interval

Random forest

5

0.736

0.573–0.898

4

0.801

0.689–0.912

4

0.797

0.527–1.066

Logistic regression

4

0.590

0.532–0.649

4

0.560

0.406–0.713

4

0.615

0.541–0.689

Support vector machine

3

0.833

0.696–0.970

2

0.835

0.767–0.903

2

0.950

0.852–1.049

Gradient boosting

6

0.782

0.683–0.882

4

0.756

0.669–0.842

3

0.885

0.755–1.015

K-nearest neighbors

2

0.668

0.379–0.957

2

0.656

0.370–0.942

1

NA

NA