2008 Poster Sessions : Insuin Resistance: Clustering and Classification

Student Name : Sangho Yoon
Advisor : Robert M. Gray
Research Areas: Information Systems
Abstract
Choosing correct models can be important to problems of statistical learning, perhaps especially to avoid over-fitting and under-fitting data. We apply various model selection algorithms to human genetics, where we define insulin resistance precisely and attempt to find predisposing patterns of polymorphisms in chosen candidate genes and other variables. Although insulin resistance is a fundamental predictor of subsequent type 2 diabetes and cardiovascular disease, it has not yet been defined precisely on the basis of measurements that devolve from current medical practice. Our sample is of Chinese women subjects who are part of SAPPHIRe. We cluster individuals based on eleven medical measurements and combine this with a bootstrap method that respects family structures to arrive at a definition. Finally, we attempt to find predisposing patterns of polymorphisms in candidate genes and other variables by mutual information and CART; and we present results based on them.

SAPPHIRe is the Stanford Asian Pacific Program in Hypertension and Insulin Resistance.

Bio
He received M.S. degrees in Electrical Engineering ('04) and Statistics ('07) from Stanford university. Currently he is a PhD candidate in Electrical Engineering at Stanford university. His research areas include pattern recognition, machine learing and data compression. His recent work is about model selection for clustering and its application to complex data such as genetics.