As @jonah suggested let’s have this post to discuss the best way to collaborate on this and whether it should be part of an existing package or a new one. As a start we could also collect models to be implemented: Royston-Parmar flexible parametric models Poisson time spliting approach (like de…

Royston-Parmar was somewhat predated by Herndon. Look at Herndon and Kooperberg references What would be an excellent approach would be to allow a prior on the shape of the hazard function, modeling the hazard as a cubic or linear spline, and to make the covariate space parametric (allowing regres…

@harrelfe Thanks for tips and references. Glad to have you in on the discussion here! Personally I haven’t done too much work in the area of survival models before, but I can advise whoever works on this (e.g. @ermeel , @kaybenleroll and others) about anything related to the software implementation.…

[image] jonah: @harrelfe Thanks for tips and references. Glad to have you in on the discussion here! Personally I haven’t done too much work in the area of survival models before, but I can advise whoever works on this (e.g. @ermeel and others) about anything related to the software implementat…

[image] ermeel: post to discuss the best way to collaborate on this and whether it should be part of an existing package or a new one I think everyone should strongly consider adding to / improving the surveHE package that is already on CRAN [image] survHE: Survival Analysis in Healt…

That’s also being discussed in the original thread: [image] Using stan_jm for parametric proportional hazards regression only rstanarm @kaybenleroll and @ermeel or anyone else interested, can you start a separate topic for this? We can discuss the best way to collaborate …

[image] harrelfe: a prior on the shape of the hazard function, modeling the hazard as a cubic or linear spline, Would you use (natural cubic) splines to model / smooth the hazard function or the (log) cumulative hazard function, like e.g. in Royston & Parmar ? The interpolation between line…

Best to use a spline on log hazard or log cumulative hazard, to force correct mathematical constraint. Royston & Parmar may be OK, just wished they had studied previous papers in the same journal. I haven’t studied the Royston & Parmar notation yet. For Kooperberg see the R polspline package and h…

Thanks will check it. While implementing Royston & Parmar, I realised that Bayesian formulation might run into problems with their choice of natural cubic splines, in particular have a look at the following [royston_problem] Here l is the factor to the likelihood which comes from a specific obser…

I wish I had the experience to speak to your question. Regarding the negative contribution to the likelihood, that seems to be a general question beyond the scope of the particular spline basis, and would welcome some input from others. On the monotonicity constraint on the cumulative hazard or l…

Survival models in rstanarm

Developers

jonah April 28, 2018, 7:56pm 7

That’s also being discussed in the original thread:

Topic		Replies	Views
Stan for survival models Modeling	26	12778	July 18, 2018
Using stan_jm for parametric proportional hazards regression only rstanarm	25	4847	May 5, 2018
Huge R-hat and low effective size in piecewise exponential model when using more than one predictor and interval Modeling fitting-issues	12	2386	September 30, 2019
Calculating hazard ratios and interpreting differences in survival and between a multi-level covariate from rstanarm::stan_surv rstanarm techniques , interpret-results , survival	3	1652	May 25, 2021
Bayesian implementation of the Royston-Parmar survival model Modeling	6	1679	November 11, 2021

Survival models in rstanarm

Related topics