search

stc04003 / aftgee

Accelerated failure time with generalized estimating equations
https://www.sychiou.com/aftgee/index.html
3 stars 4 forks source link

Support for confint and predict? #8

Open jarbet opened 1 year ago

jarbet commented 1 year ago

Confidence intervals

I see you added support for the base R confint() in v1.1.7. For some reason I am unable to install this version using devtools::install_github. I am wondering if you plan to submit this version to CRAN in the near future? I suspect I will have better luck installing from CRAN.

In the meantime, I can calculate the CIs manually.

Predictions

Would it be possible to add a function (perhaps supporting the base R predict()) to be able to predict the time of event for a new patient, given their covariate values?

I suspect this might not be possible for aftsrr since that model does not estimate the intercept, correct?

However, it should be possible to make predictions with aftgee, since this model estimates the intercept, right?

stc04003 commented 1 year ago

Thank you again for your comments. What error messages did you encounter when you tried to install v1.1.7?

You are right that the aftsrr does not estimate the intercept while aftgee can. On the other hand, it should be possible to calculate the predicted time with aftsrr, though. The package currently has a predict function for both the aftsrr and aftgee, but it seems buggy. I will look into it.

jarbet commented 1 year ago

Thank you again for your comments. What error messages did you encounter when you tried to install v1.1.7?

I was able to install v1.2.0 on my personal PC and the linux server I use for work, so I can at least use it there. However, I am unable to install the package locally on my work Mac. I get the following warnings/errors:

building ‘aftgee_1.2.0.tar.gz’ Warning: invalid uid value replaced by that for user 'nobody' Warning: invalid gid value replaced by that for user 'nobody'

warning: 'sprintf' is deprecated: This function is provided for compatibility reasons only. Due to security concerns inherent in the design of sprintf(3), it is highly recommended that you use snprintf(3) instead. [-Wdeprecated-declarations]

rankFit.cpp:224:21: warning: unused variable 'p' [-Wunused-variable] arma::uword const p = a.n_elem; ^ ld: warning: directory not found for option '-L/usr/local/gfortran/lib/gcc/x86_64-apple-darwin18/8.2.0' ld: warning: directory not found for option '-L/usr/local/gfortran/lib' ld: library not found for -lgfortran clang: error: linker command failed with exit code 1 (use -v to see invocation) make: *** [aftgee.so] Error 1 ERROR: compilation failed for package ‘aftgee’

One other Mac-user in my lab also got the above warnings/errors. Maybe Mac-users will need to install gfortran in order for it to work?

You are right that the aftsrr does not estimate the intercept while aftgee can. On the other hand, it should be possible to calculate the predicted time with aftsrr, though. The package currently has a predict function for both the aftsrr and aftgee, but it seems buggy. I will look into it.

That would be very cool to be able to make predictions for new patients with both aftsrr and aftgee! Then we could compare the prediction accuracy of these models (perhaps using C-index) versus CoxPH and other survival models (e.g. survival trees/forests). It might be possible to use your models with the tidyverse survival framework as well.