Multiple Kernel Learning with Random Effects for Predicting Longitudinal Outcomes and Data Integration

• Activity Number: 388
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Tianle Chen*+ and Donglin Zeng and Yuanjia Wang
• Companies: Columbia University and University of North Carolina at Chapel Hill and Columbia University
• Keywords: Disease prediction ; Statistical learning ; Integrative analysis ; Latent effects

Abstract

Predicting disease risk and progression is one of the main goals in clinical research. Cohort studies on the natural history and etiology of chronic diseases span years and data are collected at multiple visits. Although kernel-based statistical learning methods are proven to be powerful for a wide range of disease prediction problems, these methods are only well studied for independent data. In this paper, we develop a novel statistical learning method for longitudinal data by introducing subject-specific latent effects through a designed kernel to account for within-subject correlation. We embed our method in a multiple kernel learning framework to allow easy integration of heterogeneous data sources. We apply the developed method to two large epidemiological studies on Huntington's disease and Alzhemeier's Disease (Alzhemeier's Disease Neuroimaging Initiative, ADNI) and show substantial gain in performance while accounting for the longitudinal feature of data.