Dear ALL,

Further to the details above, I have done the following:

I imported my dataset from STATA to R. , then typed:

Library (caret)

Library (glmnet)

Library (mlbench)

Library (psych)

Library (survival)

time = PTIME

event= PSURV

X=cbind (sex, BMI_TCR, COLD_ISCH_KI, SERUM_CREAT, END_CPRA, ETHCAT ,AGE ,AMIS ,BMIS , STEROIDS_MAINT AGE_DON, DIAB, dgf )

Y=cbind ( time, event)

Coxph= coxph(Surv (time, event)~X, method “Breslow”)

Fit=glmnet (X,Y, family=”cox”)

Now , I am trying to run a LASSO inference for cox regression using all the variables in the Matrix called X.

However, I donot know how to proceed or what is the syntax to do this

I would your help

Kind regards

See our attached paper on the subject. David Booth

Check https://journal.r-project.org/archive/2017/RJ-2017-018/RJ-2017-018.pdf

Hello Ali,

Look at this

https://cran.r-project.org/doc/contrib/Faraway-PRA.pdf

Dear Chinaza and everyone.

Thank you very much for responding.

The paper about using CoxphMIC is very interesting and easy to implement.

https://journal.r-project.org/archive/2017/RJ-2017-018/RJ-2017-018.pdf

My question are-

1-After using coxphMIC and running it, it selects variables for the final model.

Can I use these selected variables only and run a cox proportional hazard model to get HR and P value and confidence intervals?

2- what are the post-estimation tests I can do after using coxphMIC?

I

Hi Ali,

did you find a solution to do LASSO inference with Cox models.

Thanks in advance,

elena