Extensions to decision curve analysis, a novel method for evaluating diagnostic tests, prediction models and molecular markers

Table 4 Confidence intervals for the net benefits using bootstrap methods.

Threshold	Net benefit	Point Estimate	Bootstrap Mean	Bootstrap confidence interval
	For:		(2000 replications)	2.5^th percentile	97.5^th percentile
15	All	0.1288	0.1288	0.0922	0.1669
	Base	0.1288	0.1290	0.0922	0.1669
	Full	0.1590	0.1613	0.1280	0.1955
	Full vs All	0.0302	0.0325	0.0172	0.0490
	Full vs Base	0.0302	0.0322	0.0170	0.0488
25	All	0.0126	0.0126	-0.0288	0.0559
	Base	0.0748	0.0735	0.0432	0.1036
	Full	0.1167	0.1214	0.0919	0.1522
	Full vs All	0.1041	0.1088	0.0802	0.1369
	Full vs Base	0.0419	0.0479	0.0257	0.0712
35	All	-0.1393	-0.1393	-0.1871	-0.0894
	Base	0.0428	0.0410	0.0181	0.0662
	Full	0.0940	0.0965	0.0652	0.1291
	Full vs All	0.2333	0.2358	0.1990	0.2712
	Full vs Base	0.0511	0.0555	0.0293	0.0836
60	All	-0.8514	-0.8513	-0.9291	-0.7703
	Base	0.0149	0.0159	-0.0041	0.0378
	Full	0.0547	0.0569	0.0331	0.0838
	Full vs All	0.9061	0.9083	0.8345	0.9824
	Full vs Base	0.0399	0.0410	0.0176	0.0676
80	All	-2.7027	-2.7026	-2.8581	-2.5405
	Base	-0.0149	-0.0028	-0.0230	0.0243
	Full	0.0189	0.0223	-0.0054	0.0527
	Full vs All	2.7216	2.7249	2.5649	2.8757
	Full vs Base	0.0338	0.0251	0.0000	0.0595

ISSN: 1472-6947