No MCMC samples for shared random effects?

[alexvolkmann]

Is it possible to access the MCMC samples for the shared random effects of a joint model?

I'm assuming that the "b" element of the "mcmc" list of a "jm" object would be the place to find them. However, this is the only element of the list that does not have the right dimensions.

I'm interested in the fitted values of the model over the different MCMC samples. In a different issue it says to construct them using the design matrices and posterior means. However, the "Mean" element of the "statistics" component of a "jm" object does not equal the "b" element, so is it just one random sample?

Thanks!

# Standard Example
pbc2.id$status2 <- as.numeric(pbc2.id$status != 'alive')
CoxFit <- coxph(Surv(years, status2) ~ sex, data = pbc2.id)
fm1 <- lme(log(serBilir) ~ year * sex, data = pbc2, random = ~ year | id)
jointFit <- jm(CoxFit, list(fm1), time_var = "year",
                n_iter = 1200L, n_burnin = 200L, n_thin = 5L)
str(jointFit$mcmc$b)

[drizopoulos]

You need to use jm(..., save_random_effects = TRUE).