How to do posterior predictive checks for survival model in interval censorship status?

Dear all,
I want to evaluate my survival model with posterior predictive checks in RSTAN.
For this purpose, I wrote the codes for generating y_rep in the log-logistic model with right-censored data as follows:
‘’’
{real u;
u=uniform_rng(0,1);
for (n in 1:N_uncensored){
y_rep[n] = ((u / (1 - u)) ^ (1 / alpha)) * ((1 / (exp ((-respon_uncensored[n,7:18] * beta - w1[county_uncensored[n]] - theta[county_uncensored[n]]) * alpha))) ^ (1 / alpha));}
}
‘’’.

But I have a problem with writing the codes to evaluate the survival model in interval censorship status. My question is how to generate the y_rep for evaluating the log-logistic survival model with interval-censored data?
Any idea will be greatly appreciated. Thank you in advance!

Regards,
Eisa

1 Like

Can anyone help me with my problem?

Any suggestions will be greatly appreciated. Thank you in advance!

Warm Regards,
Eisa

PP checks with censored dta are hard (you will notice that e.g. brms by default doesn’t let you do pp checks with censoring). Overall, you need to be able to model the censoring process and apply it to the generated y_rep. So e.g. for simple censoring, where large values are cut off and replaced with a specific bound, you need to take the y_rep and replace all large values with the bounds and then proceed with your

Does that make sense?

Best of luck with your model!

Dear Dr. Modrák,
Thanks a ton for your reply and assistance and I apologize for the delay.
Can you explain more to me about your suggestion?

Here, my data is interval-censored and my problem is in generating y_rep to this type of data in order to evaluate the model with PP checks.

Regards,
Eisa

Unfortunately, that’s not enough information to answer your question. What exactly is the process that leads to the interval censoring? Without understanding that process quite well, I don’t think you can perform PP checks.