No announcement yet.
  • Filter
  • Time
  • Show
Clear All
new posts

  • Penalized Cox regression

    Hi, does anyone know whether we can run penalized cox regression in STATA? The number of the event in my study is rather limited, and one suggestion from my professor is to use penalized likelihood for estimation. I saw examples of penalized cox regression in R, but not in STATA. Therefore, I am wondering whether STATA can handle penalized cox regression. Any advice from you will be highly appreciated. Thank you

  • #2
    I am aware that a Firth-penalized Cox regression command has been written for Stata, but to my knowledge it has never been made public.

    Have you considered a fully parametric survival regression model with judiciously chosen regularizing prior distributions on the parameters? The new bayes: prefix in Release 15 makes this more straightforward.

    I believe that at least some in the know would consider explicitly defined prior distributions to be more justifiable, especially in models with multiple predictors, than Firth penalization.


    • #3
      See this nice paper by Sullivan and Greenland


      • #4
        Dear Joseph and Andrea, thank you for your kind replies. I am new to penalized cox regression. Is there any STATA example related your suggestion?