Ordinal Rank Survival, survival with multiple outcomes

#1
Hey all -

This is mostly directed at Dr. Harrell. My PI and I were talking about survival models accounting for different endpoints in one model.

First, I'm going to present what I would have done, and may be I can get some critique.
Let's assume a parametric survival model (exponential). We have a Bayesian model, we use partial pooling (local and global prior for each patient) and then have seperate likelihood. More mathematically accurate would be to do multivaraite (i.e. a likelihood with multiple output). The first part I can implement readily in Stan, but I'm not sure the set backs because I haven't implemented it yet.

Second, I heard that Dr. Harrell suggests Ordinal Rank survival models. I would like to see this in 1) a probabilistic form 2) code. and 3) with permission, I would love to run it on some clinical data that we have access too and see what inference we can make.