American Statistical Association New York City Metropolitan Area Chapter Mailman School of Public Health Columbia University Department of Biostatistics Colloquium

Predictive models of disease based on gene expression measurements have focused on finding distinct representative profiles for healthy and diseased populations. However, some diseases, such as cancer, exhibit increased heterogeneity in the disease population. This is further complicated in gene expression, and other genomic assays, where means and variances are not independent. In this talk, we discuss how profiles (mean estimates), and heterogeneity (variability estimates) behave in cancer and their effect on predictive models.

Hector Corrada Bravo is Assistant Professor at the Center for Bioinformatics and Computational Biology and the Department of Computer Science at the University of Maryland. He works on statistical and machine learning methods for high-throughput genomic data analysis, currently focusing on second-generation sequencing and epigenetics. Previously, he was a postdoctoral fellow in the Biostatistics department of the Johns Hopkins Bloomberg School of Public Health. His research interests span the full range of computational genomics: from pre-processing of high-throughput biological assays to disease predictive and prognostic models that integrate high-throughput genomic and other data.