Skip to main content

Table 4 Confusion matrix

From: Prediction of long-term hospitalisation and all-cause mortality in patients with chronic heart failure on Dutch claims data: a machine learning approach

 

(n = 7733)

AUC*

CI

 

Sensitivity (%)

Specificity (%)

Logistic Regression

1-Year HF hospitalisation

0.7099

0.6822

0.7375

61.4

73.6

Random forest

1-Year HF hospitalisation

0.7075

0.6815

0.7335

62.3

68.1

Neural network

1-Year HF hospitalisation

0.7319

0.7061

0.7577

71.9

63.9

Elastic net

1-Year HF hospitalisation

0.7320

0.7066

0.7575

71.0

66.3

Logistic regression

3-Years HF hospitalisation

0.7255

0.7088

0.7422

71.5

62.0

Random forest

3-Years HF hospitalisation

0.7045

0.6874

0.7217

73.1

57.8

Neural network

3-Years HF hospitalisation

0.7313

0.7147

0.7479

67.3

68.9

Elastic net

3-Years HF hospitalisation

0.7330

0.7165

0.7495

67.8

67.7

Logistic regression

1-Year all-cause mortality

0.7746

0.7568

0.7923

78.0

63.4

Random forest

1-Year all-cause mortality

0.7649

0.7471

0.7827

59.8

80.1

Neural network

1-Year all-cause mortality

0.7664

0.7483

0.7845

76.7

62.6

Elastic net

1-Year all-cause mortality

0.7866

0.7691

0.8040

74.2

69.7

Logistic regression

3-Years all-cause mortality

0.7897

0.7790

0.8003

79.2

63.1

Random forest

3-Years all-cause mortality

0.7639

0.7527

0.7751

71.8

67.8

Neural network

3-Years all-cause mortality

0.7817

0.7709

0.7925

75.0

67.1

Elastic net

3-Years all-cause mortality

0.7911

0.7805

0.8017

64.4

78.1

  1. *Area under the Curve, 95% confidence interval