BMC Medical Research Methodology

Table 1 Simulation results for estimation of incident/dynamic cause specific AUC of a single marker in Scenario 1, comparing the methods cWMR, Semi-parametric and cFPL

From: Inference about time-dependent prognostic accuracy measures in the presence of competing risks

				cWMR				Semi-parametric					cFPL
Cause	Log(t)	\(\sum _{i=1}^{i=n}R_{i}(t)\)	AUC	Mean	ARB	BSE	SE	CP	Mean	ARB	BSE	CP	Mean	ARB	BSE	SE	CP
1	-1.5	407	0.833	0.830	0.36	0.031	0.030	88.0	0.815	2.16	0.046	90.0	0.833	0	0.026	0.028	93.1
	-1.2	347	0.802	0.800	0.25	0.030	0.029	89.0	0.783	2.37	0.050	89.6	0.802	0	0.025	0.027	93.5
	-0.9	272	0.771	0.771	0	0.030	0.029	84.1	0.753	2.33	0.053	89.6	0.771	0	0.026	0.028	92.1
	-0.6	192	0.743	0.743	0	0.032	0.031	87.3	0.727	2.15	0.055	93.4	0.742	0.1	0.030	0.031	91.3
	-0.3	119	0.716	0.719	0.41	0.038	0.036	88.6	0.707	1.25	0.057	90.0	0.716	0	0.036	0.038	93.1
	0.0	63	0.693	0.696	0.43	0.048	0.046	86.9	0.684	1.23	0.058	90.2	0.694	0.14	0.053	0.050	92.8
	0.3	28	0.672	0.676	0.6	0.071	0.064	89.7	0.670	0.29	0.056	89.0	0.678	0.89	0.094	0.084	93.1
	0.6	10	0.654	0.659	0.76	0.126	0.104	86.0	0.653	0.15	0.060	92.6	0.653	0.15	0.211	0.186	87.8
	0.9	4	0.639	0.657	2.91	0.189	0.169	89.0	0.578	9.54	0.119	82.1	0.599	6.10	0.353	0.586	75.6
2	-1.5	407	0.5	0.501	0.2	0.051	0.050	87.3	0.504	0.8	0.082	90.2	0.5	0	0.045	0.049	92.1
	-1.2	347	0.5	0.501	0.2	0.045	0.044	88.5	0.503	0.6	0.078	91.2	0.5	0	0.040	0.042	91.9
	-0.9	272	0.5	0.501	0.2	0.041	0.041	88.8	0.499	0.4	0.075	90.0	0.499	0.2	0.037	0.04	92.1
	-0.6	192	0.5	0.500	0.0	0.042	0.041	89.1	0.501	0.4	0.071	86.8	0.499	0.2	0.041	0.041	90.9
	-0.3	119	0.5	0.499	0.2	0.047	0.045	90.5	0.494	1.2	0.069	90.2	0.498	0.4	0.047	0.049	93.2
	0.0	63	0.5	0.499	0.2	0.057	0.054	86.3	0.496	0.8	0.065	91.4	0.498	0.4	0.068	0.061	91.3
	0.3	28	0.5	0.499	0.2	0.080	0.073	87.3	0.505	1.0	0.063	90.4	0.496	0.8	0.120	0.107	92.5
	0.6	10	0.5	0.500	0.0	0.136	0.111	83.3	0.496	1.6	0.062	89.2	0.498	0.4	0.247	0.223	87.7
	0.9	4	0.5	0.499	0.2	0.209	0.191	87.1	0.471	5.8	0.103	92.0	0.495	1.0	0.376	0.497	62.8

Average of bootstrap mean (Mean), average of bootstrap standard errors (BSE), absolute relative bias (ARB), model-based standard error (SE), coverage probability (CP)(nominal level is 90 percentage) of AUC

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com