Title: "Measurement error models with applications to biomedical sciences"
Speaker: Xiao-Feng Wang, Department of Quantitative Health Sciences/Biostatistics Cleveland Clinic Foundation; School of Medicine Case Western Reserve University
Place: Physics (PHYS) 112; February 16, 2010, Tuesday, 4:30pm

Abstract

Data measured with errors occur frequently in many biomedical fields. For example, in low level microarray data from either the cDNA microarray or the Affymetrix GeneChip system, each observation is an original signal coupled with a background noise. In neuroimage studies, observed predictors are often summarized from regions of interest and thereby involve errors-in-variables problems. Statistical analyses that ignore measurement errors can yield biased and inconsistent estimates and thus lead to erroneous conclusions with various degrees in data analysis. In this talk, we present two real medical studies in genetics and neuroscience, where different error problems take place. We address two measurement error models along with solution algorithms to correct for errors, including the Fourier-type deconvolution and the Simulation-extrapolation (SIMEX). Some simulations and real data analysis are demonstrated to illustrate the use of the methods and their relative merits.

Associated Reading:
Xiao-Feng Wang, Zhaozhi Fan, Bin Wang. 2010. Estimating smooth distribution function in th e presence of heteroscedastic measurement errors. Computational Statistics and Data Analysis 54:25-36.



Click here for a full schedule of BIOINFORMATICS SEMINARS, past and present.