Should stan_jm throw an error when the survival submodel has zero events?

[tessaclarkson]

Summary:

Getting this error attempting to run stam_jm "error occurred during calling the sampler; sampling not done Error in check_stanfit(stanfit) : Invalid stanfit object produced please report bug"

Description:

Model attempting to run: mod1 <- stan_jm(formulaLong = dsmcc_sumanxiety ~ ses + race + time + (time | id), dataLong = longd, formulaEvent = survival::Surv(max.pt, attrit) ~ ses + race, dataEvent = eventd, time_var = "time", chains = 1, refresh = 2000, seed = 12345)

Getting this Error: Chain 1: Rejecting initial value: Chain 1: Error evaluating the log probability at the initial value. Chain 1: Exception: Exception: normal_lpdf: Random variable[9] is nan, but must not be nan! (in '/functions/mvmer_functions.stan' at line 291; included from 'model_jm' at line 12) (in 'model_jm' at line 119)

Chain 1: Chain 1: Initialization between (-2, 2) failed after 100 attempts. Chain 1: Try specifying initial values, reducing ranges of constrained values, or reparameterizing the model. [1] "Error in sampler$call_sampler(args_list[[i]]) : Initialization failed." error occurred during calling the sampler; sampling not done Error in check_stanfit(stanfit) : Invalid stanfit object produced please report bug

Reproducible Steps:

If applicable, the steps required to reproduce the issue. If you have a reproducible example, please include it. Screenshot of Data for Longitudinal model:

Screenshot of Data for Event model:

tried the following: xcode-select --install result: xcode-select: error: command line tools are already installed

install.packages("rstanarm", type = "source")

library(devtools) install_github("stan-dev/rstanarm")

and none of these fixed the problem.

RStanARM Version:

The version of the rstanarm package you are running (e.g., from packageVersion("rstanarm"))

R Version:

The version of R you are running (e.g., from getRversion()) RStudio Version 1.4.1106 R Version: ‘4.0.4’

Operating System:

Catalina, OS X 10.15.7

[tessaclarkson]

following up the version of rstanarm is ‘2.21.1’

[tessaclarkson]

seems like the fix was related to miss-coding my events model to the point where it censored out all of my subjects that had events rather than noting as events! once i change this the problem was resolved!

[sambrilleman]

Awesome, thanks for reporting back, that is good to hear! Perhaps it should be catching that at the data stage tho, rather than letting it get to the model fitting stage! 🤔

If that isn't happening, you could possibly leave the issue open and change the title to something like "Should stan_jm throw an error when the survival submodel has zero events"...