Conferencias
Mauricio Castro
Universidad Concepción, Chile
Despite technological advancements in efficiency enhancement of quantification assays, biomedical studies on HIV RNA measures viral load responses that are often subjected to some detection limits. Moreover, some related covariates such as CD4 cell count may be often measured with substantial errors. Censored nonlinear mixed-effects models are routinely used to analyze this type of data and are based on normality assumptions for the random terms. However, derived inference may not be robust when the underlying normality assumptions are questionable (thick-tails). In this article, we address these issues simultaneously under a Bayesian paradigm through joint modeling of response and covariate processes using an attractive class of normal/independent (NI) densities. The NI family produces symmetric heavy-tailed distributions that includes the normal distribution, the Student-t, slash and the contaminated normal distributions as special cases. The methodology is illustrated using a case study on longitudinal HIV viral loads. Both simulation and real data analysis reveal that our models are capable of providing robust inference for heavy-tailed situations commonly encountered in HIV/AIDS, or other clinical studies. Joint work with V. Lachos and D. Bandyopadhyay.
