|
Activity Number:
|
333
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #306971 |
|
Title:
|
The LASSO-Patternsearch Algorithm and Its Application to Data from the Beaver Dam Eye Study
|
|
Author(s):
|
Weiliang Shi*+ and Grace Wahba and Kristine Lee and Ronald Klein and Barbara E.K. Klein
|
|
Companies:
|
University of Wisconsin-Madison and University of Wisconsin-Madison and University of Wisconsin-Madison and University of Wisconsin-Madison and University of Wisconsin-Madison
|
|
Address:
|
Department of Statistics, 1300 University Ave., Madison, WI, 53706,
|
|
Keywords:
|
LASSO ; GLM ; risk factor patterns ; myopia
|
|
Abstract:
|
The LASSO-Patternsearch is proposed, to identify clusters of multiple risk factors for outcomes of interest in large demographic studies. Many diseases are suspected of having multiple interacting risk factors acting in concert, and it is of interest to uncover these interactions. The method is related to H. Zhang et al, JASA99(2004), except that some variable flexibility is sacrificed to allow entertaining models with low as well as high order interactions among multiple predictors. A LASSO is used to pick out important patterns, being applied conservatively to have a high rate of retention of true patterns, while allowing some noise. Then the patterns picked by the LASSO are tested in the framework of generalized linear models to reduce the noise. The method is applied to data from the Beaver Dam Eye Study and is shown to expose physiologically interesting interacting risk.
|
