Either in clinical study or biomedical research, it is a common practice to combine multiple biomarkers to improve the overall diagnostic performance. Despite the fact there exist a large number of statistical methods for biomarker combination under binary classification, research on this topic under multi-class setting is sparse. The overall diagnostic accuracy, i.e. the sum of correct classification rates, directly measures the classification accuracy of the combined biomarkers. Hence the overall accuracy can serve as an important objective function for biomarker combination, especially when the combined biomarkers are used for the purpose of making medical diagnosis. In this paper, we address the problem of combining multiple biomarkers to directly maximize the overall diagnostic accuracy by presenting several grid search methods and derivation-based methods. A comprehensive simulation study was conducted to compare the performances of these methods. An ovarian cancer data set is analyzed in the end.
TorreLATrabertBDeSantisCE, et al.
Ovarian cancer statistics, 2018.CA Cancer J Clin2018;
68: 284–296.
5.
XiongCvan BelleGMillerJP, et al.
Measuring and estimating diagnostic accuracy when there are three ordinal diagnostic groups.Stat Med2006;
25: 1251–1273.
6.
XiongCvan BelleGMillerJP, et al.
A parametric comparison of diagnostic accuracy with three ordinal diagnostic groups.Biometrical J2007;
49: 682–693.
7.
KangLTianL.Estimation of the volume under the roc surface with three ordinal diagnostic categories. Comput Stat Data Anal2013;
62: 39–51.
8.
NakasCT.Developments in roc surface analysis and assessment of diagnostic markers in three-class classification problems. Revstat Stat J2014;
12: 43–65.
9.
LiJFineJP.Roc analysis with multiple classes and multiple tests: methodology and its application in microarray studies.Biostatistics2008;
9: 566–576.
10.
ZhangYLiJ.Combining multiple markers for multi-category classification: an roc surface approach. Aust N Z J Stat2011;
53: 63–78.
11.
NakasCTYiannoutsosCT.Ordered multiple-class roc analysis with continuous measurements.Stat Med2004;
23: 3437–3449.
12.
SuJQLiuJS.Linear combinations of multiple diagnostic markers. J Am Stat Assoc1993;
88: 1350–1355.
13.
PepeMSThompsonML.Combining diagnostic test results to increase accuracy.Biostatistics2000;
1: 123–140.
14.
McIntoshMWPepeMS.Combining several screening tests: optimality of the risk score.Biometrics2002;
58: 657–664.
15.
KangLLiuATianL.Linear combination methods to improve diagnostic/prognostic accuracy on future observations.Stat Meth Med Res2016;
25: 1359–1380.
16.
LiuCLiuAHalabiS.A min–max combination of biomarkers to improve diagnostic accuracy.Stat Med2011;
30: 2005–2014.
17.
Ma H, Halabi S and Liu A. On the use of min-max combination of biomarkers to maximize the partial area under the roccurve. J Probab Stat 2019; 2019: 1–13.
18.
GaoFXiongCYanY, et al. Estimating optimum linear combination of multiple correlated diagnostic tests at a fixed specificity with receiver operating characteristic curves. J Data Sci2008;
6: 105–123.
19.
LiuASchistermanEFZhuY.On linear combinations of biomarkers to improve diagnostic accuracy.Stat Med2005;
24: 37–47.
20.
YinJTianL.Optimal linear combinations of multiple diagnostic biomarkers based on Youden index.Stat Med2014;
33: 1426–1440.
21.
YanQBantisLEStanfordJL, et al.
Combining multiple biomarkers linearly to maximize the partial area under the roc curve.Stat Med2018;
37: 627–642.
22.
KangLXiongCCraneP, et al.
Linear combinations of biomarkers to improve diagnostic accuracy with three ordinal diagnostic categories.Stat Med2013;
32: 631–643.
23.
Maiti R, Li J, Das P, et al. A distribution-free smoothed combination method of biomarkers to improve diagnostic accuracy in multi-category classification. arXiv preprint 2019; arXiv:1904.10046.
24.
LiJFineJPPencinaMJ.Multi-category diagnostic accuracy based on logistic regression. Stat Theory Relate Fields2017;
1: 143–158.
25.
SchistermanEFPerkinsN.Confidence intervals for the Youden index and corresponding optimal cut-point. Commun Stat-Simul C2007;
36: 549–563.
26.
LiuX.Classification accuracy and cut point selection. Stat Med2012;
31: 2676–2686.
27.
AttwoodKTianLXiongC.Diagnostic thresholds with three ordinal groups.J Biopharm Stat2014;
24: 608–633.
28.
YinJTianL.Joint confidence region estimation for area under roc curve and Youden index.Stat Med2014;
33: 985–1000.
29.
HuaJTianL.A comprehensive and comparative review of optimal cut-points selection methods for diseases with multiple ordinal stages.J Biopharm Stat2020;
30: 46–68.
30.
NakasCTAlonzoTAYiannoutsosCT.Accuracy and cut-off point selection in three-class classification problems using a generalization of the Youden index. Stat Med2010;
29: 2946–2955.
31.
HsuMJChenYH.Optimal linear combination of biomarkers for multi-category diagnosis.Stat Med2016;
35: 202–213.
32.
HsuMJHsuehHM.The linear combinations of biomarkers which maximize the partial area under the roc curves. Comput Stat2013;
28: 647–666.
33.
SilvermanBW.Density estimation for statistics and data analysis.
Abingdon, UK:
Routledge, 2018.
34.
HuangXQinGFangY.Optimal combinations of diagnostic tests based on AUC.Biometrics2011;
67: 568–576.
35.
Chudecka-GłazAM.Roma, an algorithm for ovarian cancer.Clin Chim Acta2015;
440: 143–151.
36.
CramerDWBastRCBergCD, et al.
Ovarian cancer biomarker performance in prostate, lung, colorectal, and ovarian cancer screening trial specimens.Cancer Prev Res2011;
4: 365–374.
37.
TeamPPGohaganJKProrokPC, et al.
The prostate, lung, colorectal and ovarian (PLCO) cancer screening trial of the national cancer institute: history, organization, and status. Control Clin Trials2000;
21: 251S–272S.
38.
BuysSSPartridgeEGreeneMH, et al.
Ovarian cancer screening in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial: findings from the initial screen of a randomized trial.Am J Obstet Gynecol2005;
193: 1630–1639.
39.
BuysSSPartridgeEBlackA, et al.
Effect of screening on ovarian cancer mortality: the prostate, lung, colorectal and ovarian (plco) cancer screening randomized controlled trial.J Am Med Assoc2011;
305: 2295–2303.
40.
OkenMMHockingWGKvalePA, et al.
Screening by chest radiograph and lung cancer mortality: the prostate, lung, colorectal, and ovarian (PLCO) randomized trial.J Am Med Assoc2011;
306: 1865–1873.
41.
MaSHuangJ.Combining multiple markers for classification using roc.Biometrics2007;
63: 751–757.
42.
XuT, et al.
Flexible combination of multiple diagnostic biomarkers to improve diagnostic accuracy.BMC Med Res Methodol2015,
15: 94.
43.
YouyiFYinSHuangY.Combining biomarkers linearly and nonlinearly for classification using the area under the ROC curve.Stat Med2016,
35: 3792–3809.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
