B0997
Title: Bayesian scale mixture of normal censored linear mixed models with within-subject serial dependence
Authors: Fernanda Schumacher - The Ohio State University (United States) [presenting]
Kelin Zhong - University of Connecticut (United States)
Victor Hugo Lachos Davila - University of Connecticut (United States)
Abstract: HIV RNA viral load measures are often subjected to some upper or lower detection limit, depending on the quantification assays. Hence, the responses are either left- or right-censored. Censored mixed-effects models are routinely used to analyze this type of data and are based on normality assumptions for the random terms. However, those assumptions might not provide robust inference in the presence of atypical observations. A Bayesian analysis of censored linear models is developed replacing the Gaussian assumptions with the flexible class of scale mixture of normal (SMN) distributions while accounting for within-subject serial correlation through useful dependence structures and taking advantage of the No-U-Turn sampler (NUTS) to obtain posterior simulations. The SMN is an attractive class of symmetric heavy-tailed distributions that includes the normal distribution, the Student-t, slash, and the contaminated normal distributions as special cases. To illustrate the flexibility and applicability of the proposed model, an HIV AIDS study on viral loads dataset will be analyzed.