Bayesian analysis of multivariate longitudinal data using latent structures with applications to medical data

Thời gian: 15h Thứ 5, ngày 11/05/2023

Địa điểm: Phòng 507, nhà A6

Link online Zoom: 845 8621 8812

Passcode: 123456

Tóm tắt: In biomedical science, sociology, psychology, etc., it is often of interest to understand how multiple variables are associated. In many cases, the observed outcomes are not of direct interest, instead they are considered as manifestations of one or more underlying latent characteristics. When measured repeatedly over time, interest is then often placed on the evolution of the latent characteristics (variables) and/or the effects of covariates on those evolution, rather than on the observed characteristics. Compared to statistical models for multivariate observed longitudinal data, models for latent variables have been examined to a lesser extent. We then proposed new Bayesian methods to address clinical research questions arising from biomedical researchers. We choose the Bayesian approach because it is able to simultaneously estimate the latent variables and the model parameters, allowing to incorporate uncertainty in parameter estimation into latent variable estimation in a natural way. Methodologically, we have covered a wide range of settings for multivariate longitudinal latent variables in balanced and unbalanced designs. We conducted extensive simulation studies to demonstrate the advantages of our proposed models over the existing approaches. Our methods have been also successfully applied to BelRAI data (a database collected on Belgian frail individuals of age 65 or older living at home but at risk of institutionalization) and a publicly available data set involving amyotrophic lateral sclerosis (ALS) patients. But, our methods can also be applied to any dataset with a similar structure in other areas such as sociology, psychology, etc