國立政治大學統計學系主講人：許緯文助理教授(Department of Statistics, Kansas State University)
學 術 演 講
學 術 演 講
題 目：A Covariate-Adjusted Classification Model for Multiple Biomarkers in
Disease Screening and Diagnosis
時 間：民國107年5月21日 (星期一) 下午1：30
The classification methods based on linear biomarker combinations have been well established and widely used to improve the accuracy in disease screening and diagnosis. However, it is seldom to include covariates such as gender and age at diagnosis into these classification procedures. It is known that some covariates are often associated with biomarkers or patient outcomes in practice, therefore the inclusion of covariates may further increase the power of prediction and improve the classification accuracy. In this paper, a covariate-adjusted classification model for multiple biomarkers is proposed. Technically, it is a two-stage model with a parametric or non-parametric approach to combine multiple biomarkers first, and then incorporating covariates with the use of maximum rank correlation estimators. Specifically, these parameter coefficients associated with covariates can be estimated by maximizing the area under the receiver operating characteristic (ROC) curve. The asymptotic properties of these estimators in the model are also provided. An intensive simulation study is conducted to evaluate the performance in finite sample sizes, and the gene data for early detection of colorectal cancer are used to illustrate the proposed methodology.