Skip to main content

Table 1 Comparing discrimination and calibration of different models

From: Study on the prediction model of atherosclerotic cardiovascular disease in the rural Xinjiang population based on survival analysis

 

Cox regression

Lasso-Cox

RSF

China-PAR

FRS

Man

 AUC

0.789

0.787

0.791

0.757

0.729

 95% CI for the AUC

(0.765–0.812)

(0.763,0.810)

(0.767,0.813)

(0.732,0.781)

(0.723,0.773)

 C statistic

0.779

0.775

0.780

0.748

0.738

 95% CI for the C statistic

(0.736–0.822)

(0.730,0.819)

(0.730,0.829)

(0.695,0.800)

(0.688,0.789)

 Brier Score

0.061

0.062

0.060

0.063

0.064

 χ2 for Homser-Lemeshow test

10.16

4.20

10.12

49.20

24.35

 P for Homser-Lemeshow test

0.25

0.84

0.26

< 0.01

< 0.01

Woman

 AUC

0.756

0.759

0.759

0.756

0.742

 95% CI for the AUC

(0.730,0.780)

(0.733,0.783)

(0.734,0.783)

(0.730,0.780)

(0.716,0.767)

 C statistic

0.733

0.736

0.737

0.734

0.721

 95% CI for the C statistic

(0.697,0.768)

(0.701,0.771)

(0.702,0.771)

(0.699,0.768)

(0.686,0.757)

 Brier Score

0.111

0.111

0.110

0.125

0.143

 χ2 for Homser-Lemeshow test

10.84

15.22

3.72

394.87

-

 P for Homser-Lemeshow test

0.21

0.06

0.88

< 0.01

-

  1. Abbreviations: RSF random survival fores, China-PAR Prediction for ASCVD Risk in China, FRS Framingham model