Skip to main content

Table 2 Simulation results for correction for over-fit.

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

Threshold

probability

Number of events

(Total sample size)

"Best estimate"

of net benefit

Uncorrected

Correction

Method

    

Bootstrap

Repeated

2-fold

Repeated

10-fold

15%

100

(385)

0.1590

0.1618

(0.1438, 0.1799)

0.1589

(0.1406, 0.1771)

0.1569

(0.1412, 0.1734)

0.1585

(0.1425, 0.1755)

 

50

(193)

 

0.1625

(0.1327, 0.1940)

0.1558

(0.1250, 0.1878)

0.1542

(0.1268, 0.1845)

0.1577

(0.1292, 0.1887)

 

40

(154)

 

0.1625

(0.1288, 0.1972)

0.1540

(0.1181, 0.1902)

0.1517

(0.1185, 0.1877)

0.1565

(0.1222, 0.1937)

 

30

(116)

 

0.1630

(0.1233, 0.2039)

0.1510

(0.1088, 0.1943)

0.1471

(0.1093, 0.1881)

0.1545

(0.1152, 0.1974)

 

20

(77)

 

0.1670

(0.1155, 0.2224)

0.1484

(0.0946, 0.2044)

0.1341

(0.0877, 0.1837)

0.1509

(0.0998, 0.2061)

25%

100

(385)

0.1167

0.1216

(0.1021, 0.1420)

0.1172

(0.0977, 0.1381)

0.1160

(0.0988, 0.1342)

0.1174

(0.0990, 0.1363)

 

50

(193)

 

0.1234

(0.0900, 0.1585)

0.1143

(0.0802, 0.1504)

0.1129

(0.0834, 0.1460)

0.1169

(0.0856, 0.1522)

 

40

(154)

 

0.1243

(0.0861, 0.1640)

0.1131

(0.0733, 0.1541)

0.1104

(0.0747, 0.1483)

0.1155

(0.0787, 0.1556)

 

30

(116)

 

0.1241

(0.0783, 0.1709)

0.1088

(0.0601, 0.1562)

0.1058

(0.0660, 0.1500)

0.1134

(0.0717, 0.1591)

 

20

(77)

 

0.1297

(0.0741, 0.1892)

0.1069

(0.0474, 0.1679)

0.0918

(0.0411, 0.1440)

0.1098

(0.0529, 0.1712)

35%

100

(385)

0.0940

0.0967

(0.0744, 0.1189)

0.0916

(0.0692, 0.1139)

0.0893

(0.0715, 0.1087)

0.0916

(0.0720, 0.1121)

 

50

(193)

 

0.0980

(0.0652, 0.1347)

0.0877

(0.0535, 0.1251)

0.0857

(0.0557, 0.1198)

0.0901

(0.0574, 0.1260)

 

40

(154)

 

0.0993

(0.0602, 0.1411)

0.0865

(0.0455, 0.1297)

0.0829

(0.0456, 0.1212)

0.0883

(0.0495, 0.1281)

 

30

(116)

 

0.0999

(0.0531, 0.1482)

0.0829

(0.0334, 0.1316)

0.0782

(0.0359, 0.1237)

0.0856

(0.0407, 0.1348)

 

20

(77)

 

0.1048

(0.0472, 0.1665)

0.0795

(0.0178, 0.1421)

0.0636

(0.0134, 0.1162)

0.0822

(0.0238, 0.1436)

60%

100

(385)

0.0547

0.0568

(0.0377, 0.0765)

0.0498

(0.0302, 0.0699)

0.0446

(0.0275, 0.0628)

0.0495

(0.0320, 0.0688)

 

50

(193)

 

0.0574

(0.0236, 0.0922)

0.0433

(0.0083, 0.0793)

0.0362

(0.0052, 0.0701)

0.0441

(0.0103, 0.0792)

 

40

(154)

 

0.0593

(0.0181, 0.1007)

0.0421

(-0.0001, 0.0851)

0.0316

(-0.0045, 0.0708)

0.0405

(-0.0013, 0.0831)

 

30

(116)

 

0.0608

(0.0123, 0.1116)

0.0384

(-0.0119, 0.0915)

0.0255

(-0.0152, 0.0715)

0.0373

(-0.0107, 0.0889)

 

20

(77)

 

0.0650

(0.0060, 0.1286)

0.0324

(-0.0289, 0.0981)

0.0082

(-0.0392, 0.0610)

0.0291

(-0.0321, 0.0976)

80%

100

(385)

0.0189

0.0218

(0.0000, 0.0477)

0.0140

(-0.0068, 0.0391)

0.0104

(-0.0064, 0.0305)

0.0139

(-0.0032, 0.0350)

 

50

(193)

 

0.0259

(-0.0148, 0.0670)

0.0100

(-0.0305, 0.0535)

0.0008

(-0.0305, 0.0377)

0.0100

(-0.0237, 0.0521)

 

40

(154)

 

0.0287

(-0.0178, 0.0779)

0.0088

(-0.0381, 0.0581)

-0.0061

(-0.0424, 0.0357)

0.0068

(-0.0360, 0.0537)

 

30

(116)

 

0.0351

(-0.0160, 0.0924)

0.0091

(-0.0436, 0.0638)

-0.0160

(-0.0611, 0.0354)

0.0022

(-0.0492, 0.0597)

 

20

(77)

 

0.0404

(-0.0227, 0.1077)

0.0010

(-0.0672, 0.0714)

-0.0434

(-0.1057, 0.0199)

-0.0106

(-0.0872, 0.0678)

  1. Values are mean (2.5th – 97.5th centile) of 2000 replications.
Back to article page

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us