Model-Based Clustering of Longitudinal Data: Application to Modeling Disease Course and Gene Expression Trajectories

We consider the problem of clustering time-dependent data. The model is a mixture of regressions, with variance–covariance matrices that are allowed to vary within the extended linear mixed model family. We discuss applications to biomedical data and analyze two longitudinal data sets: one on patients with delirium, and the other on mosquito gene expression following infection.