From:"Monder, Harvey"   
Subject:[NMusers] NONMEM vs NLMIXED  
Date:Wed, March 13, 2002 12:56 pm

In view of the difficulty in validating NONMEM, some individuals have
suggested that the new NLMIXED procedure in SAS may be a viable alternative
as SAS is a validated/validatable product.  I have not seen this procedure
nor have I seen any references to results produced by it.  Has anyone had
any experience with this software?  What is the functionality as compared to
NONMEM?

        Regards,
        Harvey


*******

From:"Amir A. Tahami"   
Subject:Re: [NMusers] NONMEM vs NLMIXED  
Date:Wed, March 13, 2002 2:39 pm  


Dear Harvey;

        NLINMIX is a SAS macro for fitting nonlinear mixed models using
PROC NLIN and PROC MIXED. You can find this SAS macro (program written
in SAS) and more details about that at the following link:

http://ewe3.sas.com/techsup/download/stat/nlmm800.html

Best regards,
Amir A. TAHAMI

 


*******


From:"Nandy, Partha"   
Subject:RE: [NMusers] NONMEM vs NLMIXED  
Date:Wed, March 13, 2002 2:41 pm  


Hi,

Following up on the issue of validation..Does anyone have experience with
validation of NONMEM and if so, to what extent the software was validated?

I will appreciate the group's input in this regard.

Thanks
Partha

 


*******

From:"Kowalski, Ken"   
Subject:RE: [NMusers] NONMEM vs NLMIXED  
Date:Wed, March 13, 2002 2:55 pm  

Amir, Harvey, NMUSERS:

Please note that NLMIXED (Harvey's original question) is different from
NLINMIX.  NLMIXED is a full-fledge PROC available in SAS Version 8.  It has
much the same functionality as NONMEM's $PRED for models that you want to
specify on your own, however, it has nothing comparable to PREDPP.  NLMIXED
does have options for exact maximum likelihood estimation using
quadrature-based methods in addition to first-order approximation methods
such as those in NONMEM.

Ken