Regularization Methods for Predicting an Ordinal Response Using Longitudinal High-Dimensional Genomic Data

• Activity Number: 466
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Jiayi Hou*+ and Kellie Jo Archer
• Companies: University of California, San Diego and Virginia Commonwealth University
• Keywords: classification ; high-dimensional data ; longitudinal data ; ordinal response ; regularization methods

Abstract

An ordinal scale is commonly used to measure health status and disease related outcomes in hospital settings as well as in translational research. In addition, repeated measurements are common in clinical practice for tracking and monitoring the progression of complex diseases. Classical methodology based on statistical inference has contributed to the analysis of data in which the response categories are ordered and the number of covariates remains smaller than the sample size. With the emergence of genomic technologies being increasingly applied for more accurate diagnosis and prognosis, a novel high-dimensional data where the number of covariates is much larger than the number of samples are generated. To meet the emerging needs, we introduce our proposed model which consists of two parts: extend the GMIFS method to ordinal model; and combine the GMIFS procedure with the classical mixed-effects model to construct an innovative penalized random coefficient ordinal response model for classifying the disease progression along with time. We demonstrate the accuracy of the proposed model in classification using a time-course microarray dataset collected from a burn injury study.