A1201
Title: A Bayesian semiparametric model for survival analysis with a right censored covariate
Authors: Farouk Nathoo - University of Victoria (Canada) [presenting]
Puneet Velidi - University of Victoria (Canada)
Maryclare Griffin - University of Massachusetts Amherst (United States)
Tanya Garcia - University of North Carolina at Chapel Hill (United States)
Roland Matsouaka - Duke University (United States)
Sujit Ghosh - North Carolina State University (United States)
Juxin Liu - University of Saskatchewan (Canada)
Jing Qian - University of Massachusetts, Amherst (United States)
Abstract: Survival models incorporating censored covariates are useful for understanding Huntington's disease progression because subject observation often ends before the initial presentation of symptoms. Modeling both a censored response and censored covariate is challenging due to ordering constraints (e.g., ordering of disease stages) leading to non-independent censoring mechanisms. We develop SPARTACCUS, a Bayesian joint model with semiparametric Cox proportional hazards models for both the right-censored response and censored covariate, including uncensored covariates. A computationally efficient MCMC sampler is implemented that uses data augmentation to impute censored values using posterior predictive distribution. In simulations, SPARTACCUS outperforms complete case analysis in high-censoring settings across a range of sample sizes and hazard shapes. Applying our method to the PREDICT-HD clinical trial shows that increased baseline parietal lobe grey matter volume is associated with increased hazard of Stage II Huntington's, accounting for time to Stage I symptoms. This finding may inform the development of non-invasive brain stimulation therapies for Huntington's.
