A0239
Title: Deep mixture of linear mixed models for complex longitudinal data
Authors: Lucas Kock - National University of Singapore (Singapore) [presenting]
Nadja Klein - Karlsruhe Institute of Technology (Germany)
David Nott - National University of Singapore (Singapore)
Abstract: Mixtures of linear mixed models are widely used for modeling longitudinal data for which observation times differ between subjects. In typical applications, temporal trends are described using a basis expansion, with basis coefficients treated as random effects varying by subject. A key advantage of these models is that they provide a natural mechanism for clustering, which can be helpful for interpretation. Current versions of mixtures of linear mixed models are not specifically designed for cases where there are many observations per subject and a complex temporal trend, which requires a large number of basis functions to capture. In this case, the subject-specific basis coefficients are a high-dimensional random effects vector, for which the covariance matrix is hard to specify and estimate, especially if it varies between mixture components. To address this issue, we consider the use of recently developed deep mixture of factor analyzers models as the prior for the random effects. The resulting deep mixture of linear mixed models is well-suited to high-dimensional settings. We demonstrate the adaptability of our approach across a range of real-world applications, including within-subject prediction for an unbalanced longitudinal study, the task of predictive likelihood-free inference as well as missing data imputation.
