Fack, a cox model I'm working on is warning me on infinite estimates. I'm going to follow @carlislerainey.bsky.social 's advice and report the coxph fit with infinite estimates and the coxphf fit, but was wondering if there is a simple implementation of coxph in stan - ThreadSky

About ThreadSky

phdemetri.bsky.social • 5 days ago

Fack, a cox model I'm working on is warning me on infinite estimates.

I'm going to follow @carlislerainey.bsky.social 's advice and report the coxph fit with infinite estimates and the coxphf fit, but was wondering if there is a simple implementation of coxph in stan

Comments

phdemetri.bsky.social•5 days ago

Please, do not make me write this myself

leobelzile.bsky.social•5 days ago

The hazard is fitted via splines, so in practice easier to view this via the Bayesian lense.

leobelzile.bsky.social•5 days ago

If you are willing to go away from the Cox model (by replacing the nonparametric hazard with a semiparametric component), you can use the life table approach. You bin the time axis and run a Poisson regression: see https://doi.org/10.2307/2346901, https://doi.org/10.2307/2986301, https://doi.org/10.1080/01621459.1981.10477634

phdemetri.bsky.social•5 days ago

I’ve done something similar to this, but I’d rather not introduce complexity for the other authors (who are physicians). Much easier to say “Bayesian cox” than “no no it’s really the same, believe me!”

jordannafa.bsky.social•5 days ago

brms has a cox model, AFAIK

mpopov.com•5 days ago

Yep, family `cox()` supports log, identity, and softplus for the proportional hazards parameter.

gfgm.bsky.social•5 days ago

There's a nice discussion in the stan user manual here https://mc-stan.org/docs/stan-users-guide/survival.html#proportional-hazards-model although my impression is that it's not that simple

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply