Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 172 - Prediction and Misclassification in Biomedical Research
Type: Contributed
Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
Sponsor: Section on Statistics in Epidemiology
Abstract #313393
Title: A Placement Value-Based ROC Surface Regression Approach to Evaluate Diagnostic Capacity of a Biomarker in Predicting Diseases with Three Categories
Author(s): Soutik Ghosal* and Katherine Grantz and Zhen Chen
Companies: Eunice Kennedy Shriver National Institute of Child Health and Human Development and Eunice Kennedy Shriver National Institutes of Child Health and Human Development and NICHD/NIH
Keywords: ROC surface; diagnostic accuracy modeling; biomarker data analysis; obstetrics; abnormal birthweights

Receiver operating characteristics (ROC) curve analysis has been a popular tool in diagnostic accuracy modeling to evaluate the prognostic capacity of a biomarker in predicting binary disease outcomes. However, for disease outcomes with more than two categories, this framework becomes ineffective and a more general framework of manifold becomes necessary. We propose a placement value-based ROC surface regression approach in the presence of an ordinal outcome with three categories. The use of placement value is to facilitate a direct assessment of covariate effect on the measure of diagnostic accuracy. Simulation studies are provided to assess the performance of the proposed model under various scenarios. To illustrate the model’s applicability, we analyze the NICHD Fetal Growth Study data to estimate the diagnostic ability of Estimated fetal weight (EFW), an ultrasound biomarker in predicting abnormal birth categories: small-for-gestational-age (SGA), appropriate-for-gestational-age (AGA), and large-for-gestational-age (LGA).

Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program