A0338
Title: Mediation analysis of community context effects on heart failure using the survival R2D2 prior
Authors: Brandon Feng - North Carolina State University (United States) [presenting]
Eric Yanchenko - Akita International University (Japan)
Lloyd Hill - United States Environmental Protection Agency (United States)
Lindsey Rosman - University of North Carolina at Chapel Hill (United States)
Brian Reich - North Carolina State University (United States)
Ana Rappold - United States Environmental Protection Agency (United States)
Abstract: Congestive heart failure (CHF) is a leading cause of morbidity, mortality, and healthcare costs, impacting >23 million individuals worldwide. The vast amount of time-varying health data collected by electronic health records (EHR) provides a critical opportunity to improve risk stratification and clinical management of CHF, yet statistical inference on large amounts of EHR data is still challenging. Thus, accurately identifying influential risk factors is pivotal to reducing information dimensionality. Bayesian variable selection in survival regression is a common approach to solving this problem. The aim is to propose placing a beta prior directly on the model coefficient of determination (Bayesian R2), which induces a prior on the global variance of the predictors and provides shrinkage. Through reparameterization using an auxiliary variable, a majority of the parameters are updated with Gibbs sampling, simplifying computation and quickening convergence. Performance gains over competing variable selection methods are showcased through an extensive simulation study. Finally, the method is applied in a mediation analysis to identify community contextual factors that accelerate the time to incident CHF diagnosis of patients with preexisting cardiovascular disease. The model has high predictive performance, and it is found that factors associated with higher socioeconomic inequality increase the risk of heart failure.
