Activity Number:
|
320
- Methods Tailored to Unique Data and Trial Features
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Section on Medical Devices and Diagnostics
|
Abstract #313864
|
|
Title:
|
A Regression Approach to Estimating the Discrete Diagnostic Likelihood Ratio for Nontraditional Biomarkers
|
Author(s):
|
Hanna Lindner* and Warren Bilker and Phyllis Gimotty
|
Companies:
|
University of Pennsylvania and University of Pennsylvania and University of Pennsylvania
|
Keywords:
|
multinomial regression;
biomarker;
discrimination
|
Abstract:
|
The diagnostic likelihood ratio (DLR) is a popular epidemiologic measure of discrimination and predictive ability of a diagnostic test. The area under the receiver operating characteristic (ROC) curve (AUC) is a more common statistic for measuring discrimination, however it assumes a monotone relationship between risk of disease and the test. The discrete and continuous DLRs overcome this limitation in that they do not assume a monotone relationship. This feature of DLR can be used to assess the discriminatory ability of nontraditional biomarkers, which are biomarkers where both excessively low and high values are associated with disease resulting in a nonmonotone relationship between the risk of disease and the biomarker. We propose a flexible regression method that allows for estimation of the discrete DLR, and assessment of covariate effects on the discrete DLR. This is particularly useful to identify patient characteristics that modify a biomarkers ability to discriminate between those with and without disease. The proposed methods are illustrated with a data application.
|
Authors who are presenting talks have a * after their name.