Measures to Evaluate Biomarkers as Predictors of Incident Cases and Connection to Information Theory

• Activity Number: 333
• Type: Invited
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Medical Devices and Diagnostics
• Author(s): Patrick Heagerty* and Jason Liang
• Companies: University of Washington and University of Washington

Abstract

In many biomedical applications a primary goal is to predict incident or future cases and appropriate measures that characterize predictive potential or incremental value are needed. We first review non-parametric methods proposed for incident time-dependent accuracy (Zheng and Heagerty, 2005; Saha and Heagerty, 2013) and then overview extensions of integrated discrimination index (IDI) that are appropriate for hazard models. New methods are connected to information theory based criteria for model choice. Non-parametric estimation is outlined and application to benchmark data sets is presented to illustrate the methodology.