Prediction metrics for survival analysis

FWIW, I recently came across the SurvMetrics R package that implements several prediction accuracy metrics for time-to-event/survival analysis.

Perhaps it would be possible to incorporate some of these metrics into the tidymodels survival framework?

Hi! Thanks for the suggestion! We are currently working on adding performance metrics for survival analysis. The ones first on our list are the Brier score, the integrated Brier score, the concordance index, and ROC AUC. You can follow the progress in the development version of yardstick!

1 Like

I will also say that we are very close to getting the whole of tidymodels working with event time data (to the point where I've written the draft blog post).

In the next month or so, we'll do a cascade of releases with all of the functionality and guides on how to use it.

2 Likes

Thanks Hannah and Max for your replies! I am Watching the Yardstick github repo now and look forward to learning more about using tidymodels to analyze time-to-event data!

This topic was automatically closed 21 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.