Activity Number:

251
 SPEED: Biopharmaceutical Methods and Application I, Part 2

Type:

Contributed

Date/Time:

Monday, July 29, 2019 : 2:00 PM to 2:45 PM

Sponsor:

Biopharmaceutical Section

Abstract #307612


Title:

The Application of Beta Regression for Modeling a Covariate Adjusted ROC

Author(s):

Xing Meng* and Jack D. Tubbs

Companies:

Baylor University and Baylor University

Keywords:

Placement values;
beta regression;
ROC regression

Abstract:

The receiver operating characteristic (ROC) curve illustrates the diagnostic capabilities of the binary classifier system. In many applications, the test performance is affected by covariates. A solution to this problem can be found by modeling the ROC as a function of the covariate effects using a generalized linear model framework. There are three common models that can be used to estimate a covariateadjusted ROC. These include a parametric, semiparametric and beta regression model. The beta regression model has been shown to perform very well using the placement values, the probability that the test result of the diseased subject exceeds the test result of the nondiseased subject with the same covariate value. The beta regression model also avoids the correlation problem found in the parametric and semiparametric methods. This paper will use data from a clinic study to compare the performance of the three models.
