volfovsky
commented
3 years ago Currently that’s not an option but you can consider a manual implementation of a predictive error function such as in the PE_functions.R file. We are working on streamlining this process.
One thing to note is that causal inference for time-to-event data is notoriously hard so there could be other considerations that must be addressed to make the AME framework meaningful.
Best, Alex
Alexander Volfovsky Assistant Professor of Statistical Science Co-Director of the Polarization Lab Duke University volfovsky.github.iohttp://volfovsky.github.io On Apr 12, 2021, 4:30 PM -0400, Matthew Loop @.***>, wrote:
The Cox proportional hazards model is ubiquitous in medical research. Especially if your sample size is small, you want to take advantage of a continuous (albeit censored) time to event outcome vs. a binary outcome.
Could the user specify family = "cox", which would be passed to the glmnet call to determine whether the covariates that are matched on predict the outcome?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https://github.com/vittorioorlandi/FLAME/issues/2__;!!OToaGQ!7voRcdDat-DK4Faq4OsmPfAhG-IC_WynyuSuzeZwjVfSng8KvdsY7FKhJse-dn-hm6I_aTQ3-yU$, or unsubscribehttps://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AC362B2FPQFQB2MAZYXKE6TTINJ7PANCNFSM42Z7ZYGQ__;!!OToaGQ!7voRcdDat-DK4Faq4OsmPfAhG-IC_WynyuSuzeZwjVfSng8KvdsY7FKhJse-dn-hm6I_60mE-Qo$.
The Cox proportional hazards model is ubiquitous in medical research. Especially if your sample size is small, you want to take advantage of a continuous (albeit censored) time to event outcome vs. a binary outcome.
Could the user specify
family = "cox", which would be passed to theglmnetcall to determine whether the covariates that are matched on predict the outcome?