Classification of Distinct Trajectories in Longitudinal Data with Irregularly Spaced Intervals: A Large Data Application of Post-Hoc Mixture Modeling of BLUPs from Mixed Models

• Activity Number: 132 - SLDS CSpeed 1
• Type: Contributed
• Date/Time: Monday, August 9, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Md Jobayer Hossain* and Benjamin E Leiby
• Companies: Nemours Children's Health System and Thomas Jefferson University
• Keywords: Classification; longitudinal trajectories; Irregularly-spaced; mixture-based; mixed-effects models; Gaussian finite mixture

Abstract

Classification of longitudinal trajectories in naturally occurring large data is an area of growing interest. While a few approaches including recently introduced deep learning methods have shown utility for clustering longitudinal data with fixed time, only mixture-based mixed-effects models implemented in R package lcmm and M-plus are mainly used for longitudinal data with irregularly-spaced interval. Both packages are affected by computational complexities, especially in large datasets. In recent applications, our newly introduced post-hoc mixture modeling of BLUPs (PMMB) from mixed-effects models have shown better performance than the above mixture-based approaches. We have shown that the application of post-hoc Gaussian finite mixtures on empirical BLUPs can aptly identify the presence of subgroups. It assumes a mixture of subgroups with normally distributed random effects of varying geometric features that are determined by the eigenvalue decomposition of the covariance matrix. This study would apply PMMB to a large dataset of early childhood growth patterns and evaluate the classification performance through simulations and model-based external validations.