|
Activity Number:
|
607
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #304212 |
|
Title:
|
Adjusting for Matched Samples in Discriminant Analysis of Schizophrenia Biomarkers for Post-Mortem Tissue Studies
|
|
Author(s):
|
Josephine Asafu-Adjei*+ and Allan Sampson and Robert Sweet
|
|
Companies:
|
University of Pittsburgh and University of Pittsburgh and University of Pittsburgh
|
|
Address:
|
2717 Cathedral of Learning , Pittsburgh, PA, 15260,
|
|
Keywords:
|
linear discrimination analysis ; CART ; matched design ; post-mortem studies ; schizophrenia
|
|
Abstract:
|
In post-mortem tissue studies for schizophrenia, control and schizophrenia subjects are pair matched on many variables and measured on a set of covariates. In the literature, results are integrated across studies to identify biomarkers that can distinguish between diagnostic groups, accounting for multiple studies done on the same subject. The Stanley Foundation Neuropathology Consortium determined such biomarkers using linear discriminant analysis (LDA) and classification and regression trees (CART). We present an improved approach to these types of analyses. Specifically, we provide an approach applied to LDA and CART that accounts for study design and covariates. This method is applied to a post-mortem schizophrenia tissue data set and it is shown that the new analyses produce classification results superior to those previously used.
|
