A Bayesian Cox Proportional-Hazards Regression with Functional Covariates

• Activity Number: 243
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Eunjee Lee*+ and Joseph Ibrahim and Hongtu Zhu
• Companies: and University of North Carolina and University of North Carolina at Chapel Hill
• Keywords: Functional Linear Model ; Bayesian Survival Analysis ; Functional Principal Component Analysis ; Cox Proportional-Hazards Regression ; ADNI

Abstract

Functional linear model has been got a lot of attention to incorporate functional covariates in the linear model. Bayesian survival model with functional covariates, however has not been well studied. We propose a Bayesian Cox proportional-hazards regression for survival data with functional covariates. Functional principal component analysis is employed to take into account the effects of the functional covariates in our model. The coefficients are estimated by Markov chain Monte Carlo methods. Simulation studies and an analysis of a real imaging data set from the Alzheimer's Disease Neuroimaging Initiative (ADNI) are used to illustrate our model. In Alzheimer's Disease (AD), it is critical to predict the timing of conversion to a Mild Cognitive Impairment (MCI) patient to the AD. Our main interest is examining the integration of imaging, clinical, and genetic information for predicting the time-to-conversion outcomes. To assess the overall model fitting, we will check the deviance residual and risk scores plots.