On Model Building for Combining Genomics and Clinical Covariates Information

• Activity Number: 80
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 4:00 PM to 5:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Samir Lababidi*+
• Companies: Food & Drug Administration
• Keywords: Clinical Trial ; gene expression ; clinical utility ; model building ; genomic classifier

Abstract

The early papers in microarray gene expression promised better prediction in cancer outcome classification than what have been done before for prognostic and diagnostic medicine using clinical information only. So far, several attempts have been made to show that the combined information based on clinical covariates and the genomic data may have better classification outcomes. However, the actual gain in the predictive ability due to the use of gene expression classifiers can depend on how these two sources of information are combined in the model building process. Here in this talk, we will present different strategies for combining the genomics and the clinical information and compare the combined model to the clinical-covariate-only model using data from the recently published studies in the MicroArray Quality Control-Phase II (MAQC-II) Project.