|
Abstract:
|
Recent technological advancements in biomedical devices are allowing researchers in neuromuscular field to evaluate potential biomarkers as a possible replacement for conventional physical functioning scores or quality of life measures which are prone to various biases. These promising biomarkers are obtained repeatedly at each visit as an infinite dimensional functional response vector generally using a very high sampling rate. In this manuscript, we propose a functional mixed-effects model using a state space approach to understand the trajectories while imposing a smoothing spline structure on functional response at each visit and taking into account of with-in subject correlations of these curves along the longitudinal measurements. Our proposed modeling approach allows us to simultaneously a) adjust for baseline variables, b) differentiate the longitudinal changes in the smooth functional response, and c) estimate the subject and subject-time specific deviations from the population-averaged response curves. We illustrate our approach on an electrical impedance myography data which is being assessed to measure the progression of Duchenne muscular dystrophy non-invasively.
|
