Re: [Psn-general] SSE question

[Kajsa H. <kaj...@fa...>]

Hello Ethan,

Is it possible to create the alternative model with two $PROB, one  
with IGNORE and MSFO=msf and the other without IGNORE, with REWIND,  
with $MSFI=msf and MAXEVAL=0 and $TABLE IPRED ...? I do not know if  
NONMEM will accept that.

In sse you might have to set option -prepend_model_file_name so that  
tables with IPRED from different samples get different names when  
copied to m1 directory.

sse version 3.1.7 and later will accept two $PROB and handle numbering  
of the msf-files so that different samples do not interfere with each  
other. Statistics will only be computed for the first $PROB. The  
output of second $PROB will be be ignored in sse_results.csv, but you  
would get the IPRED in $TABLE and ofv in raw_results.csv.

Best regards,
Kajsa

Quoting Ethan Wu <eth...@ya...>:

> Hi all,
>  I am running a simulation-estimation using sse in PsN to check a  
> shorter study
> design. The alternative model to be tested on the simulated data is  
> actually the
> same model as the original model, but use IGNORE option to cut short the
> sampling schedule.
>
>  Not only I would like to check the model parameter estimate, which PsN would
> calculate for me. I would like to check how the model prediction for  
> those later
> ignored sampling time (using IGNORE option). Of course I have to do  
> a lot post
> processing to get that, and one of thing was that, since I used  
> IGNORE option,
> the alternative model would not have those later time points in the dataset,
> therefore, I would not get IPRED for those time in the output table. I was
> wondering if there is way that I could directly get predicted value  
> (IPRED) for
> those skipped later time point by the alternative model, instead  
> that I have to
> pool out the model and simulated datasets, and take out IGNORE option, and do
> MAXEVL=0.
>  Thank you.
>