All Times ET
Keywords: Classification, Longitudinal, Mixture, Mixed effects
Recognizing distinct trajectories, identifying potential contributors to heterogeneity, and making data-driven decisions are areas of growing interest in the analysis of longitudinal data in this era of precision medicine and data science. A vast amount of naturally occurring data can be used to derive evidence-based knowledge for facilitating precise diagnosis, prevention, and tailored treatment. Mixed effects models with a mixture distribution of random effects classify trajectories through estimating profiles and computing posterior probabilities of belonging to a class. Methods are available in R and M-plus among standard packages. While theoretically sound, the method doesn’t fully address the problem mainly due to computational complexities. Linear mixed effects models ballpark curves of irregularly spaced longitudinal data with optimum precision. Random effects (REs), assumed to be distributed as multivariate normal (MVN), vary across individuals accounting for sources of heterogeneity. With presence of subgroups and assuming a mixture of MVN of REs, application of post-hoc Gaussian finite mixtures (GMM) on empirical BLUPs can classify trajectories. Resulting ellipsoids can vary by center or by other features such as size, volume and orientation that are determined by eigenvalue decomposition of covariance matrix of the REs. This study evaluates the classification performance of this new method and compares it with that of existing methods using real and simulated data of varying overlapping levels.