Table 3.

Performance Of Different Combination Methods Using The 11 Non-Invasive Score Systems In The Testing Population (n = 1,644)

	AUC	Sensitivity	Specificity	+LR	−LR	+PV	−PV	Youden Index	P-value
Majority voting	0.699 (0.670,0.728)	0.637 (0.588,0.685)	0.668 (0.641,0.694)	1.920 (1.723,2.139)	0.543 (0.474,0.622)	0.379 (0.342,0.417)	0.853 (0.829,0.874)	0.305	0
Weighted voting	0.701 (0.672,0.730)	0.637 (0.588,0.685)	0.668 (0.641,0.694)	1.920 (1.723,2.139)	0.543 (0.474,0.622)	0.379 (0.342,0.417)	0.853 (0.829,0.874)	0.305	0
Majority voting with model selection	0.802 (0.780,0.825)	0.662 (0.614,0.709)	0.702 (0.676,0.728)	2.227 (1.994,2.487)	0.480 (0.417,0.554)	0.415 (0.376,0.454)	0.867 (0.845,0.888)	0.364	0
Weighted voting with model selection	0.802 (0.780,0.825)	0.662 (0.614,0.709)	0.702 (0.676,0.728)	2.227 (1.994,2.487)	0.480 (0.417,0.554)	0.415 (0.376,0.454)	0.867 (0.845,0.888)	0.364	0
Stacking: Logistic regression	0.698 (0.669,0.728)	0.627 (0.578,0.675)	0.676 (0.649,0.702)	1.936 (1.734,2.179)	0.551 (0.483,0.630)	0.381 (0.344,0.420)	0.851 (0.822,0.867)	0.303	0
Stacking: LASSO	0.699 (0.670,0.723)	0.627 (0.578,0.675)	0.676 (0.649,0.702)	1.936 (1.734,2.179)	0.551 (0.483,0.630)	0.381 (0.344,0.420)	0.851 (0.822,0.867)	0.303	0
Stacking: SCAD	0.699 (0.670,0.723)	0.627 (0.578,0.675)	0.676 (0.649,0.702)	1.936 (1.734,2.162)	0.551 (0.483,0.630)	0.382 (0.344,0.421)	0.845 (0.827,0.872)	0.303	0
Stacking: MCP	0.699 (0.670,0.723)	0.627 (0.578,0.675)	0.676 (0.649,0.702)	1.936 (1.734,2.162)	0.551 (0.483,0.630)	0.382 (0.344,0.421)	0.845 (0.827,0.872)	0.303	0
Stacking: Stepwise regression	0.699 (0.670,0.723)	0.627 (0.578,0.675)	0.676 (0.649,0.702)	1.936 (1.734,2.162)	0.551 (0.483,0.630)	0.382 (0.344,0.421)	0.845 (0.827,0.872)	0.303	0

Notes: Indicated in parentheses for AUC, sensitivity, specificity, positive and negative likelihood ratios, and positive and negative predictive values are the corresponding 95% confidence intervals.