主 题：Jackknife Empirical Likelihood Confidence Regions for theEvaluation of Continuous-scale Diagnostic Tests with Verification Bias
主讲人：Prof. Gengsheng (Jeff) Qin
主持人：林华珍教授
时 间：2015年7月9日下午5点-6点
地 点：通博楼B212学术会议室
主办单位：统计研究中心 统计学院 科研处
主讲人简介：
Gengsheng (Jeff) Qin 系美国Georgia State University数学与统计系教授，Open Journal of Statistics、Statistics Research Letters 等期刊编委，1994年荣获The third best statistical research work。 荣获资助项目16项，累计资助金额140万美元，发表高水平的学术论文60余篇。
内容提要：
Recently, Wang and Qin (2013) proposed various bias-correctedempirical likelihood confidence regions for any two of the three parameters,sensitivity, specificity, and cut-off value, with the remaining parameter fixedat a given value in the evaluation of a continuous-scale diagnostic test withverification bias. In order to apply those methods, quantiles of the limitingweighted chi-squared distributions of the empirical log-likelihood ratiostatistics should be estimated. In order to facilitate application and reducecomputation burden, in this paper, jackknife empirical likelihood-based methodsare proposed for any pairs of sensitivity, specificity and cut-off value, andasymptotic results can be derived accordingly. The proposed methods can beeasily implemented to construct confidence regions for the evaluation ofcontinuous-scale diagnostic tests with verification bias. Simulation studiesare conducted to evaluate the finite sample performance and robustness of theproposed jackknife empirical likelihood-based confidence regions in terms ofcoverage probabilities. Finally, a real example is provided to illustrate theapplication of new methods.
