3 replies to this topic

[zhoumeng]

I have a small plasma concentration dataset with 13 potential covariates (e.g., body weight, age, sex, breed, etc). I have selected all 13 potential covariates in a one-compartment PK model and I am using Phoenix NLME to run a "Cov. Srch. Stepwise", but it shows "No unfrozen covariate effects are defined. Run cancelled".  Does anyone know how to fix this?

 

Thanks,

Zhoumeng

[mittyright]

Dear Zhoumeng,

 

could you please provide the text model ?

I suspect the covariate effects are not added to the structural parameters, please check covariate settings in Parameters Tab -> Structural Sub-Tab

Please also look into the Covariate modeling with the PK01 dataset chapter in the User's Guide.

 

BR, 

Mittyright

[Simon Davis]

Hi Lin, (and Mittyright ;0).

 

  the problem I think is quite simple, although you have told Phoenix you are potentially interested in these covariates, none of them have actually been selected by toggling to Yes from No on these boxes.

 

  That said I would be wary in just running all possible covariates, one it will take some time and two you may end up with significant but not clinically relevant covariates, especially if you number of subjects is relatively small.  I would propose to review the covariate plots and look for any that show trends, i.e the line is *not* level and in the middle like in my screenshot. That indicates etas are randomly distributed around that covariate and so it is not likely to be significant

 

  Simon

 

Please also look at taking one of our online or public training courses which have special rates for academics as that will really help you make the most of Phoenix.

 

e.g. (CL-101) Introduction to Population Modeling Methodology using Phoenix NLME

http://www.certarauniversity.com/lms/

Attached Thumbnails

• 

[joybaker]

Similar question so follow this post. I have this message as well:

 

"No unfrozen covariate effects are defined. Run cancelled"

 

The difference is I select all the covariate-parameter combinations, but the covariate is IOV.

So IOV cannot be run in covariate stepwise mode?

 

 

Hi Lin, (and Mittyright ;0).

 

  the problem I think is quite simple, although you have told Phoenix you are potentially interested in these covariates, none of them have actually been selected by toggling to Yes from No on these boxes.

 

  That said I would be wary in just running all possible covariates, one it will take some time and two you may end up with significant but not clinically relevant covariates, especially if you number of subjects is relatively small.  I would propose to review the covariate plots and look for any that show trends, i.e the line is *not* level and in the middle like in my screenshot. That indicates etas are randomly distributed around that covariate and so it is not likely to be significant

 

  Simon

 

Please also look at taking one of our online or public training courses which have special rates for academics as that will really help you make the most of Phoenix.

 

e.g. (CL-101) Introduction to Population Modeling Methodology using Phoenix NLME

http://www.certarauniversity.com/lms/

 

Attached Thumbnails

•