Abstract:
|
Longitudinal modeling of lung function in DMD is complicated by a mixture of both growth and decline in lung function within each subject, an unknown point of separation between these phases (change point), and significant heterogeneity between individual trajectories. Linear mixed-effects models can be used, assuming a fixed change point for all cases; however, this assumption may be incorrect. This paper describes an extension of linear mixed effects modeling in which random change points are integrated into the model as parameters and estimated using a generalized EM algorithm. We find that use of this "mixture modeling" approach improves the fit significantly. Substantively, modeling the change point correctly improves our understanding of the role of height in predicting lung capacity. A further result is that individual variation is more accurately modeled. This more nuanced model can be used as a prognostic indicator in DMD.
|