Modelling Error 3

Phoenix Certara Forums Phoenix WNL basics
1

I was trying for an example data on Quantitative pharmacodynamic model by Indirect response model.

I found the following error

Modelling Error 3

Variance - Covariance matrix is not of full rank or is ill-conditioned. Parameters e estimates and their associated standard errors should be interpreted with caution.

I am attaching Phoenix file along with this. Kindly help me and please check whether there were any mistakes in code and help me how to rectify the error.

PK-PD MODEL COMBINATION.phxproj (313 KB)

2

Dear Santosh

Your model is way too much overparametrized.

In addition you need to shift to Phoenix model.

Now with respect to the model you often if not always cannot estimate at the same time Emax and EC50 if you have only dose and one patient. You need therefore to assume a value either for Emax or Ec50. You put at the same time effect on production and elimination. These are most likely confounded as you see just the global effect that can be explained by either effect on production or elimination only (not both).

I am in the plane right now and will try if possible to work on this model in the plane. I will be back in Israel on Wednesday. Remind me then to help you if you did not get an answer yet.

The lesson to learn first is never use a complex model if the data do not support its use.

the error you get is just the reflection of overparametrization.

best Regards

Serge

3

santosh_serge_input.phxproj (330 KB)[quote=“minhaj mohd, username:minhaj”]

Dear Santosh

Your model is way too much overparametrized.

In addition you need to shift to Phoenix model.

Now with respect to the model you often if not always cannot estimate at the same time Emax and EC50 if you have only dose and one patient. You need therefore to assume a value either for Emax or Ec50. You put at the same time effect on production and elimination. These are most likely confounded as you see just the global effect that can be explained by either effect on production or elimination only (not both).

I am in the plane right now and will try if possible to work on this model in the plane. I will be back in Israel on Wednesday. Remind me then to help you if you did not get an answer yet.

The lesson to learn first is never use a complex model if the data do not support its use.

the error you get is just the reflection of overparametrization.

best Regards

Serge

[/quote]

Dear Santosh

please fond attached the translation of your model in ML. Note that I froze Imax to 1 as if not you will face identifiability issue. The fit returns error code 2 which is good. The se in % were reasonable with of course quite large % error for Emax and EC50 but still acceptable.

Let me know if you need further clarifications. You need to shift to Phoenix and also upgrade t 1.4.

This version is 1.3 to enable you to open the project.

best Regards

Serge

4

Thank you very much sir. The values are well in agreement with the published data values when it is run in Phoenix platform.