Hello
I want to fit a JM in rstanarm. But my longitudinal data distribution is not a default one. I want to include the distribution without changing other components of the stan_jm( ). How should I proceed?
Any suggestion would be great favor.
Thanks!
M.Alam
Hi, I am not sure this would be totally easy, but you should be able to fork rstanarm and add the distribution in glm_lp (defined here): https://github.com/stan-dev/rstanarm/blob/32f49d112f1cfc66e0a8d3dbd4bb3cf72e90743b/src/stan_files/functions/mvmer_functions.stan
You would also need to modify append_mvmer_famlink in https://github.com/stan-dev/rstanarm/blob/1d25a0c0b07ccf6c3d2cc16fc32405f08f1688e7/R/jm_data_block.R and possibly other places.
Unfortunately, it is hard to extract a simple Stan code from rstanarm so if modifying rstanarm is out of question you would need to reimplement the model from scratch (or try to find someone who has implemented it and start with their implementation).
Hope that helps.