4 replies to this topic

[Lisa Liao1412845455]

I am wondering the correct way to simulate different oral dosing regimens given just IV bolus data plus structural model. Is there somewhere we input bioavailability? I guess this wouldn't matter if all parameters are expressed without the "F" term?

 

Also, why does st error and CV% keep disappearing on THETA tab after executing the poppk analysis several times? I'm guessing it has to do with me updating the initial estimates tab.. [file name=QuinidineNov2012.phxproj size=5576396]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/QuinidineNov2012.phxproj[/file]

Attached Files

•  QuinidineNov2012.phxproj   5.32MB   0 downloads

[serge guzy]

Dear Lisa

In order to simulate extravascular with bioavailibility, you need to shift to edit graphical, then click on the absoprtion compartment. You will see down the word bioavail. On the right you will see an empty window, write for example F.

Then click on an empty space inside the graphical area, right click and select insert /parameter.

A block inside the graphical area is created. Now down, you will see also on the left an empty window (under population). Write F . Note that NLME is case sensitive.

I attached the version that include F with edit graphical.

Abnout se, I am not sure I udnerstand the issue. If it is a simulaiton, there will be no standard errors. Are you talking about a fit?

best

Serge [file name=quinidine_official_built.phxproj size=6207075]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/quinidine_official_built.phxproj[/file]

Attached Files

•  quinidine_official_built.phxproj   5.92MB   356 downloads

[Lisa Liao1412845455]

Great! The st error disappears in the 2 compartment poppk fitting...

 

Also, is there a way to graph observed concentration over 95% CI for simulated data for model validation purposes? We usually do it in R but was wondering if we can do it in NLME directly..

[serge guzy]

Dear Lisa

If the se disappar, it means that the proglram had problems inverting the hessian matrix which happens from time to time. Try to ruan the same model with QRPEM as engine and let me know if you get se.

 

now, about the seocnd request. Run your mdoel first with the simple run option (which is what you did).

then copy the model to the workflow and use the paste option to begt acopy of your model.

Accept all fixed an random (parameters/fixed effect, then you will see down that option).

now go to run option, write 0 iterations and select the last run option (vpred checks/simulation).

Then on the right, vhoose for example 100 replicates, then select the binning otpion of your time are slightly different acroos the patient population and run that mdoel.

the default will give then in the resuls the 5%, 50% and 95% confidence bounds with the observed data superimposed. If you have still an isseu, send me an example and I will do it for you.

best

Serge

PS: SORRY FOR THE TYPOS BUT NO SPELLING CHECK HERE

[Simon Davis]

Hi Lisa, just for clarity, if you see no SE in your Theta worksheet during a 'normal' fitting then if you review the Core status, around line 138 you will see the following

Starting std error computation

Warning: Hessian based standard error
computation failed in Cholesky factorization step

Sometimes this can be resolved b refining initial estimates, residual error model etc. or as Serge suggested an alternative algorithm

Simon.