In clinical practice,finding suitable linear combination of the multiple biomarkers is a commonly used technique to improve the diagnostic accuracy over a single biomarker.In this paper,we propose jackknife empirical likelihood(JEL)methods for the best linear combination of biomarkers based on the area under the receiver operating characteristic curve(AUC),and we also give the statistical inference to the combination parameters.We prove that the JEL ratio converges to the standard chi-squared distribution.Extensive simulation studies and real data analysis are also conducted,and show encouraging results. |