Estimating Biological Age Using Ensemble-Based Prediction Models

• Activity Number: 84
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Wendy Shih*+ and Steve Horvath
• Companies: and University of California, Los Angeles
• Keywords: biomarkers of aging ; genomics

Abstract

Biological age (BA) as opposed to chronological age (CA) is meant to measure the true age of an individual. The hope is that BA is superior to CA when it comes to predicting mortality or age related functional decline. Here we evaluate several statistical methods for estimating BA based on biomarkers of age. In particular, we evaluate whether ensemble based models are valuable when it comes to estimating BA based on genomic data. We compare biologic age predictions based on a multivariate regression model with those from a random generalized linear model predictor (Song et al 2013 PMID: 23323760). Further, we evaluate and adapt techniques from Klemera and Doubal (2006, PMID: 16318865) who proposed a new approach to the concept and computation of BA. Apart from simulation studies, we report the results from several genomic data applications.