Survival analysis assistance - COVID19 Research

Names: Nick Giangreco
From: Columbia University

I’m tasked with many prediction tasks that involve survival analysis in which I’m learning about but I’m new to. I’m hoping to get some expert help on asking the right questions and answering them the right way, as well as using Rmarkdown and continuous integration in the best way.

I’m proficient in R but I’m still learning all the new tidymodels package features

Maybe it would be better if you try to split your request into more specific questions, please take a look at this guides about how to more effectively ask questions here

1 Like

For an overview of survival analysis in R, you can check out the CRAN View page (or whatever it's called now), here.

For more specific questions about model choice and the like, you might try posting to Cross Validated.


Thank you @ulfelder! I’ve read the task view and a lot of the associated material. As andresrcs suggested, I think a lot of my questions could be addressed in cross validated either the programming or statistics “version”. I also have questions that I’m not sure how to ask or things I don’t know or understand fully, and that can be addressed by going through more documentation. I really appreciate y’all’s and Rstudio’s effort to help with our covid19 research. I’m currently doing these projects in collaboration with different medical departments at Columbia medical and so the questions I’m asking and insights I’m trying to derive would, I would hope, be actionable in the medical practice. I wanted to post mainly because I saw the announcement on LinkedIn and wanted to see if I could learn more from smarter people than I :slight_smile: how can I best ask or get help here versus alternate platforms like cross validated?

1 Like

Hi @ngiangre -- I'd be happy to point you to some learning resources for tidymodels, but the survival analysis features are pretty limited for now (they're being developed still) and we don't currently have any learning materials specifically focused on doing survival analysis with tidymodels. If you're able to provide a bit more information about the specific problems you're wanting to address, I can provide more concrete gudance or would be happy to loop in folks who work on related packages.


I just wanted to chime in and say that one of my favorite resources for getting started with survival analysis is this excellent tutorial by @zabore. :100:

@mine is correct that survival analysis features are currently limited and under development in tidymodels.


throwing in my 2 cents.

I would stick with the survival package that comes in base R as you are a beginner.
It's tried-and-true and contains the foundational analysis and modeling.
I've done some complex predictive analytics building from that package.
Honestly, I think more time will be spent learning survival analysis then learning how to implement them in R ... that was my experience.

I know this is an Rstudio forum and the infer package is expressive but there's still value to learning base R and other ecosystems.
You can still use tidyverse for the data transforms if you like.

Thank you all! Really appreciate the help and support. I've been able to read up enough on the statistics and implementation to use cox and aalen models within my custom prediction methodology. So right now I think I'll keep trying things out and reading these resources. But this is good to know that I'm using the right package - I thought I was missing something especially with all the new developments. And @julia this tutorial looks so great thank you so much for sharing! I think where I'm at right now in my analyses is going from the population estimation to the patient level insights from the derived models. I think I'll be able to figure it out but I have to say thank you so much for the quick responses and resources y'all! Once I have more specific questions I'll reply again.

This topic was automatically closed 42 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.