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

Table 3

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 uniform 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	1.434	0.440	0.906	0.234	0.847	0.003	0.067	0.067	0.715–0.978
15 : 15	S	1.402	0.474	1.130	0.292	0.810	0.040	0.068	0.084	0.677–0.943
	N	—	—	—	—	0.837	0.013	0.073	0.077	0.694–0.980
	P	1.418	0.338	0.907	0.181	0.847	0.003	0.052	0.050	0.745–0.950
25 : 25	S	1.373	0.353	1.060	0.212	0.820	0.030	0.053	0.062	0.716–0.923
	N	—	—	—	—	0.835	0.015	0.057	0.056	0.723–0.947
	P	1.413	0.238	0.902	0.128	0.850	0.000	0.037	0.036	0.777–0.922
50 : 50	S	1.357	0.242	0.992	0.140	0.829	0.021	0.038	0.045	0.755–0.902
	N	—	—	—	—	0.835	0.015	0.040	0.044	0.756–0.915
	P	1.402	0.167	0.908	0.090	0.850	0.000	0.026	0.024	0.798–0.901
100 : 100	S	1.350	0.169	0.967	0.097	0.832	0.018	0.027	0.033	0.780–0.885
	N	—	—	—	—	0.834	0.016	0.029	0.032	0.778–0.891

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