|
Activity Number:
|
569
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305469 |
|
Title:
|
Multiple Imputation in Right-Truncated Multivariate Normal Distribution, with Applications to RT-PCR
|
|
Author(s):
|
Abhijit Dasgupta*+ and Terry Hysolp and Scott Waldman
|
|
Companies:
|
Thomas Jefferson University and Thomas Jefferson University and Thomas Jefferson University
|
|
Address:
|
Division of Biostatistics, Philadelphia, PA, 19107,
|
|
Keywords:
|
multiple imputation ; right truncation ; multivariate distributions ; RT-PCR
|
|
Abstract:
|
Data from biomarker studies can be left- or right-truncated due to limitations of the assay. Efficiency-adjusted estimates of relative quantification of RT-PCR data are often right-truncated due to a limit in the number of cycles run and low copy number in the sample. When multiple genes from the same sample are run using RT-PCR, we can observe right-truncation in the estimates on several variables. We propose a multivariate method of imputing the missing data in the right tail of the distribution of the estimates, and show that this method is less biased in estimating the parameters of the underlying distribution and results in better test properties than imputing the truncated data separately for each marginal univariate right-truncated distribution.
|
