|
Activity Number:
|
274
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #302372 |
|
Title:
|
Using Bernstein Polynomials To Model Misclassification in BI-RADS Breast Density Measurements
|
|
Author(s):
|
Charlotte C. Gard*+ and Elizabeth R. Brown
|
|
Companies:
|
University of Washington and University of Washington
|
|
Address:
|
Department of Biostatistics, Seattle, WA, 98195,
|
|
Keywords:
|
breast density ; misclassification ; Bernstein polynomials
|
|
Abstract:
|
Breast cancer risk prediction models that include breast density (BD) measured on the four-point American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) scale have been shown to have slightly better predictive accuracy than models that do not include BD. Because BI-RADS BD is subject to misclassification, it is thought that inclusion of BD measured continuously could further improve the predictive accuracy of the models. Unlike BI-RADS BD, however, continuous BD is not routinely measured in clinical practice. With a final goal of predicting continuous BD for women for whom it is not available, we take a Bayesian approach to modeling the misclassification in BI-RADS BD. Continuous BD is modeled using Bernstein polynomial priors. We allow for radiologist-specific misclassification in BI-RADS BD and linkage of this misclassification to radiologist characteristics.
|
