Jump to content


Model Assistance (Visual Predictive Check for Plasma with 2 Compounds)

pharmacokinetics pharmacodynamics pk/pd population pharmacokinetics

  • Please log in to reply
3 replies to this topic

#1 csheme


    Advanced Member

  • Members
  • PipPipPip
  • 59 posts

Posted 12 May 2017 - 11:05 PM

Hi all,


If anyone has the time to open my attached model (linked below), I am having trouble with the visual predictive check for Plasma. There are 2 compounds (same type of model is being used to compare both drugs).


Could you please help me with my model to get an acceptable Visual Predictive Check for plasma. The Confidence interval bands for the lower interval are not coming up, is there a way to cheat the system due to negative observations.


If you are able to view my model, maybe you can tell me your opinion of the model diagnostics and goodness of fit plots, do they seem acceptable to you, maybe there is a problem with the type of error (multiplicative) that is giving rise to these negative plasma observations in the VPC?


Please assist with my model if possible.


Thanks greatly,



Edited by csheme, 12 May 2017 - 11:24 PM.

#2 serge guzy

serge guzy

    Advanced Member

  • Members
  • PipPipPip
  • 485 posts

Posted 13 May 2017 - 07:06 AM

Just change the multiplicative error model and put 0.01 instead of whatever you had estimated.

I tried to attach the project corrected but it did not work.

I will try again. Also change CEps to CEps1, CEps2 and CEPs3. Tp be correct, you cannot have the sane name. You would see that if you would shift to edti as texutal.

best Regard


#3 csheme


    Advanced Member

  • Members
  • PipPipPip
  • 59 posts

Posted 15 May 2017 - 10:44 PM

Thanks Serge, I did apply your advice and now the VPC is working.

#4 kmohan2



  • Members
  • Pip
  • 1 posts

Posted 21 November 2019 - 11:48 PM


I have a similar problem. In the VPC that I did I first had an Additive error model and the Simulated 5% quantile concentrations were negative.


Then I followed your advice above changed the error model to Multiplicative and set the Stddev = 0.01. I was able to get rid off the negative values in 5% Quantile simulation.


Dose it mean I have to redo the population PK parameter estimation with a Multiplicative error model and re-estimate stddev ? I originally used an additive error model to estimate the population PK parameters. 


I would greatly appreciate any advice here. Thanks!

Also tagged with one or more of these keywords: pharmacokinetics, pharmacodynamics, pk/pd, population pharmacokinetics

4 user(s) are reading this topic

0 members, 4 guests, 0 anonymous users