Comparison of Semiparametric, Parametric, and Nonparametric ROC Analysis for Continuous Diagnostic Tests Using a Simulation Study and Acute Coronary Syndrome Data

Table 1

Means of the parameter estimates and AUC’s with their standard errors, RMSE, and 95% confidence intervals (CI) obtained from the parametric, semiparametric, and nonparametric ROC methods using various sample sizes from 1000 simulated data sets generated from the normal distribution.

		Parameters
		𝑎=1.400,𝑏=0.900 and 𝐴𝑈𝐶=0.850
n 1 : n 0	Methods	𝑎	𝑆𝐸(𝑎)	𝑏̂	𝑆𝐸(𝑏̂)	𝐴𝑈𝐶	BIAS for 𝐴𝑈𝐶	SE   𝐴𝑈𝐶	RMSE for 𝐴𝑈𝐶	95% CI for 𝐴𝑈𝐶
	P	1.465	0.451	0.939	0.242	0.844	0.006	0.067	0.069	0.713–0.976
15 : 15	S	1.578	0.507	1.170	0.302	0.822	0.028	0.072	0.075	0.681–0.963
	N	—	—	—	—	0.849	0.001	0.070	0.071	0.712–0.986
	P	1.432	0.341	0.911	0.182	0.847	0.003	0.052	0.053	0.745–0.949
25 : 25	S	1.446	0.356	1.007	0.202	0.834	0.016	0.052	0.058	0.731–0.937
	N	—	—	—	—	0.850	0.000	0.054	0.054	0.743–0.956
	P	1.429	0.240	0.914	0.129	0.850	0.000	0.037	0.037	0.778–0.922
50 : 50	S	1.430	0.245	0.966	0.137	0.843	0.007	0.037	0.039	0.771–0.915
	N	—	—	—	—	0.852	0.002	0.038	0.038	0.776–0.927
	P	1.411	0.168	0.905	0.091	0.850	0.000	0.026	0.027	0.799–0.901
100 : 100	S	1.409	0.170	0.932	0.093	0.846	0.004	0.026	0.027	0.795–0.897
	N	—	—	—	—	0.851	0.001	0.027	0.027	0.797–0.904

P: parametric, S: semiparametric, N: nonparametric.