Error in PK modelling

Hello I am trying to perform PK modeling for a drug. I performed an NCA and got the PK parameters.

I have four columns of data treatment, patient ID, time and conc.

I’m getting the following the following error- Failed to fit model

*** NOTE *** A weight could not be computed for a data observation
Weight set equal to zero

*** ERROR 10203 *** AN ERROR OCCURRED DURING CURVE STRIPPING.
*** ERROR 10201 *** INITIAL ESTIMATES CANNOT BE DETERMINED FOR THIS MODEL.

Could you help? Should I map patient ID and treatment to Sort?
I tried 1st order one compartment

Can you share the project and data? Without that it is difficult to know what is going on.

What error model did you use? If multiplicative, do you have zero observed values? Remove those if you have them.

Best is to send the project for review.

best Regards

Serge

Meenu - you appear to be using the WNL classic engine, and probably your data is insufficient for the WNL classic engine to work out it’s own initial esitmates. have any of you subjects fitted? if so simply use their final estimates as initial estimates for the other subjects.

Simon.

Thank you Dr Serge and Dr Simon! I could run it finally after sorting by “patient ID”. But I am getting different condition numbers for each patient’s PK parameters. How can I compare between different models? I am currently trying 1 compartment extravascular. Which condition number will I use to compare between models? In the diagnostics section, again I am getiing values for each patient. How do i compare between models for the best fit?

Meenu, If you are modelling the data as individuals, I would probably look at comparing summaries of AIC (or SBC), remembering that lower numbers are better, when comparing the nested models on the SAME data.

However you should also be considering the plot as well as precision of parameters obtained (ie. CV%) as well as the planned use of the model for future simulations etc.

If you can post the project we can perhaps make some more specific suggestions.

SImon.

PK.phxproj (516 KB)

Dear Sir,

Thank you so much for your reply. I am posting the project here for your kind perusal.

Can you also tell me how to get a pooled diagnostic (and plot) to compare my model? Also, which one’s are important to be considered while comparing?

Sir I posted the project in the above thread. Could you please have a look?

Thank you so much!

Dear Meenu,

I would suggest to switch to Phoenix NLME model, Naive Pooled engine could be used with WNL license. Then you can select exactly the same model (see the NLME User’s guide, ‘An example of individual modeling with Phoenix model’, but you need to switch to population mode). Create another model object(s) and Model Comparer object too.

Regarding diagnostics: no simple answer. Try this one:

http://onlinelibrary.wiley.com/doi/10.1002/psp4.12161/pdf

BR,

Mittyright

PS: please plan to visit NLME courses (online or not) if you’d like to go deeper

PK_for_Meenu.phxproj (1.17 MB)