Jump to content


Photo

How are theta tab CI calculated?


  • Please log in to reply
2 replies to this topic

#1 enixon

enixon

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 03 October 2022 - 12:32 PM

Hi, 

 

I've written up the results of some modeling work and I have been asked to add how the parameter estimate CI are calculated. We expected it would be 1.96xSE, but that doesn't appear to be the case. I cannot find the information anywhere in the help files. Someone suggested they are likely derived using the estimate covariance matrix assuming normality. 

 

Can anyone confirm how the parameter estimate CI in the theta tab are calculated using Phoenix NLME?

 

Best regards,
Emma 



#2 smouksassi1

smouksassi1

    Advanced Member

  • Members
  • PipPipPip
  • 231 posts
  • LocationMontreal

Posted 03 October 2022 - 01:06 PM

can you share  your theta sheet 

 

tried on a random project on my end and it is all good:

 
Parameter Estimate Units Stderr   CV%          2.5% CI   97.5% CI
tvKa          0.38615         0.02771   7.175942   0.33183   0.44047
 
see redo the computation in R you get exactly the same values
 
tvKa <- 0.3861499
seka <- 0.027709891
100*seka/tvKa
 
tvKa + 1.96*seka
tvKa - 1.96*seka
0.33183 0.4404698

Edited by smouksassi1, 03 October 2022 - 01:07 PM.


#3 enixon

enixon

    Newbie

  • Members
  • Pip
  • 2 posts

Posted 04 October 2022 - 01:03 PM

I can't share the whole theta sheet but can give an example parameter:
 
Parameter    Estimate    Units   Stderr     CV%           2.5% CI     97.5% CI
tvV                2767.255   ml       246.125  8.894195    2263.876   3270.633

 

If I use 1.96xSE then:

 

2.5% CI      97.5% CI

2284.849    3249.660

 

The model estimates the human PK parameters and allometrically scales those back to fit the observed animal PK data- could this be the cause?


Edited by enixon, 04 October 2022 - 01:03 PM.





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users