I actually thought about re-scaling/centering data. But I can only center age, ppfev1 and ppdlco. I will scale that and see whether it changes anything.
Yes, you are correct. My model is a random intercept model with a Brownian motion stochastic process. Random intercept is used to describe between-subject variation and Brownian motion is used to capture the evolution of outcome variable through time within each subject.
In biostatistics we commonly use BM and related stochastic processes to capture the within-subject variations. According to a preliminary study I did using the r package “lmenssp”, integrated Ornstein Uhlenbeck (IOU); which is related to scaled BM captures, within-subject variation well. So my plan was to use a more generalized version of BM, which is fractional BM for this data.
Yes, you are correct. BM and fractional BM are both (rather can be) gaussian processes. I will try to use the links you attached for my model. Thank you! :)