Jump to content


Photo

User Model with phoenix WNL


  • Please log in to reply
5 replies to this topic

#1 Aline FUCHS

Aline FUCHS

    Newbie

  • Members
  • Pip
  • 9 posts

Posted 10 May 2010 - 01:22 PM

Dear all,

 

As a PK student, I am currently using Phoenix WinNonLin in order to achieve a PK/PD analysis. Focusing for now on the PK data, I would like to write my own models, using differential equations. I am not very familiar in writing user model with phoenix (although I have done it in WNL before).

 

To get an idea on how the new user model is running and what Phoenix Modeling Language (PML) looks like, I tried to launch a simple PK analysis (1 compartment, no lag time, first order absorption, which is equivalent to Model PK 3 in the WinNonLin Library, since this model is working with my data set).

 

Here are my questions (Just note that it is NOT a population analyses, but an individual one) and let me know if I am wrong:

 

1. To run a PK analysis using a model from the library, I have exported my data set *.xls. But it seems that now, for Phoenix Model Object, data set must look more like a NonMem data set (with “.” for separation, ## to identify column header…). Should I create a new data set, following all these rules, in order to perform a user model and save it as *.dat, *.csv, *.txt?

 

If Yes:

- What should I do with my BLQ values? Should I replace it by “.” for phoenix to be able to use it? Or should I create a MDV column, even if it is not a population analysis?

- How to import this new data set in the right way to Phoenix?

 

 

2. To get my structural model, I used built-in and either chose all my parameters needed, or chose set WNL model (model 3). Afterwards, if I want to edit it as textual, a column “Aa rate” is added in Main set up and Dosing set up. Why is the set up different in these two cases (built-in and edit as textual)? How to use “Aa rate” in a subcutaneous situation?

 

3. Last point, when I try to launch this user model, I have this error message: “Could not merge sort times. Some data that is expected to be numeric is not. Please check your source data”. I can not figure out what this error means, since all my data are numeric, except “.” and column header.

 

I would be very grateful if anyone could put me on the right track.

 

Thank you,

 

Aline,



#2 Samer Mouksassi

Samer Mouksassi

    Advanced Member

  • Members
  • PipPipPip
  • 90 posts

Posted 10 May 2010 - 01:32 PM

Ho Aline,

1. To run a PK analysis using a model from the library, I have exported my data set *.xls. But it seems that now, for Phoenix Model Object, data set must look more like a NonMem data set (with “.” for separation, ## to identify column header…). Should I create a new data set, following all these rules, in order to perform a user model and save it as *.dat, *.csv, *.txt?

You can import any dataset type into phoenix.
Phoenix Models will use the imported data from phoenix regardless of the source format.
The data setup is very flexible it can have dosing in the same data as PK or in a separate dosing sheet.


If Yes:
- What should I do with my BLQ values? Should I replace it by “.” for phoenix to be able to use it? Or should I create a MDV column, even if it is not a population analysis?
- How to import this new data set in the right way to Phoenix?

BLQ can be dealt with in several ways:
Ignoing them by having MDV=1 anv value = blank or . ( make sure to select that you have MDV in inputs)
It is hard to say how to import your data without having a look. Note Phoenix has a lot of data management functionality so you can do the changes within phoenix.


2. To get my structural model, I used built-in and either chose all my parameters needed, or chose set WNL model (model 3). Afterwards, if I want to edit it as textual, a column “Aa rate” is added in Main set up and Dosing set up. Why is the set up different in these two cases (built-in and edit as textual)? How to use “Aa rate” in a subcutaneous situation?

Aa rate will be optional and required to model zero-order absorption or if you have SC infusion.


3. Last point, when I try to launch this user model, I have this error message: “Could not merge sort times. Some data that is expected to be numeric is not. Please check your source data”. I can not figure out what this error means, since all my data are numeric, except “.” and column header.

What is the format of you time ? can you copy paste the first few rows of you data.

Samer

#3 Aline FUCHS

Aline FUCHS

    Newbie

  • Members
  • Pip
  • 9 posts

Posted 10 May 2010 - 03:39 PM

Hi Samour,

 

Thank you to answer me.

 

Although, I do not understand how are defined data files in “Modeling Language Reference Guide, Chapter 3”? Any change is needed to use my data set in Phoenix Model Object?

 

Does it mean that it is possible to tick nothing in the Aa rate column if, for now, I assume a 1st order input for a subcutaneous injection, when I am in “edit as textual” case?

 

Please find attached first few rows of my data set (2 examples: the first one is used with a Model PK from the library with its dosing worksheet, the second was created to use the Phoenix model object). I tried both of them; no one is working when launched with Phoenix model object.

 

Aline [file name=PML_test.xls size=22016]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/PML_test.xls[/file]



#4 Simon Davis

Simon Davis

    Advanced Member

  • Administrators
  • 1,316 posts

Posted 10 May 2010 - 05:30 PM

Aline, I took a quick look at your Excel sheet. If you have chosen to put both sampling and dosing dates in a single sheet then make sure that you only enter dose amounts on the time records that correspond to actual doses.

And yes you could leave Rate empty and perhaps model SC with a lag instead.

Simon.

25-26 May Trial Simulator Hands-On v2.2.1, Cary, NC
8-10 June Introduction to Phoenix WinNonlin, Woburn, MA
To see a full list of Pharsight's 2010 training schedule please visit: http://www.pharsight...aining_home.php Posted Image

#5 Samer Mouksassi

Samer Mouksassi

    Advanced Member

  • Members
  • PipPipPip
  • 90 posts

Posted 10 May 2010 - 05:34 PM

it will be helpful to have your *.phxproj

 

attached is an example: AlineEXAMPLE.phxproj

Note the first time point does not make sense with SC administraton

How do we get such high concentration at T=0 ? [file name=AlineEXAMPLE.phxproj size=350119]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/AlineEXAMPLE.phxproj[/file]

 

0 0 5.02

2 2 0.91

4 4 BLQ



#6 Aline FUCHS

Aline FUCHS

    Newbie

  • Members
  • Pip
  • 9 posts

Posted 10 May 2010 - 08:45 PM

Thank you very much for your answers.

 

 The example was definitely useful. I have been able to successfully launch this simple model and I hope to write my own models. (Value at T=0 hr is logical since it is an endogenous compound).

 

Finally, two last questions:

 

Is it possible to weight the values or should I write it in the model?

 

Is it possible to assign each subject their own initial parameter estimates (using textual model) even if I run the model simultaneously for all subject, by sorting them?

 

Aline






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users