Jump to content


Photo

(De)convolution: IV 2compt, extravascular 1 compt. how to translate?

IVIVC deconvolution convolution

  • Please log in to reply
4 replies to this topic

#1 Vanissl

Vanissl

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 17 April 2020 - 01:46 PM

Hello everyone,

 

As of my understanding for numerical deconvolution or convolution the first step was to fit a model to the IV Bolus to obtain the UIR/Macroconstants that need to be corrected by the dose.

These UIR will then be used in the convolution with my input rate to obtain a drug profile of an extravascular administration.

 

In the case that the IV Bolus follows a 2 compartment body model (Macroconstants: A,B, alpha, beta), but the extravascular adminstration has to be 1 compartment body model, is it correct to assume that I am only interested in the last part of the IV Bolus function which shows the actual elimination (the inital phase of the IV shows distribution phase, that i dont have in the extravascular adminstration).

 

To be more concrete:

1) IV Bolus (2 compt) gives me A,B, alpha, beta

2) to obtain convoluted Extravacsular profile (1 compt) I only use B and beta as unit impulse response  

 

Is that correct?

 

Thank you very much for your guidance!

 

Best wishes



#2 Simon Davis

Simon Davis

    Advanced Member

  • Administrators
  • 1,116 posts

Posted 17 April 2020 - 02:10 PM

this is a good question, and the reason is fairly common in that the the distribution phase is 'masked' in Absorpton process.  However I think IVIVC should take account of this.

Can you perhaps try both approaches and see which models your circumstances better.

 

 Simon.



#3 Vanissl

Vanissl

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 17 April 2020 - 02:17 PM

Dear Simon,

 

Thank you very much for this prompt reply!

 

I have tried both approaches and I get a perfect fit when i disregard the distribution phase (only use B and beta), whereas I greatly overestimate the cmax when I integrate the whole function.

 

Best,

Simon



#4 Simon Davis

Simon Davis

    Advanced Member

  • Administrators
  • 1,116 posts

Posted 17 April 2020 - 02:37 PM

Well I think that confirms your approach.  Remember a major purpose of modelling is to describe the observed data well..

 

rather than trying to make data 'fit' a model!

have a good weeke end.  Simon.



#5 Vanissl

Vanissl

    Newbie

  • Members
  • Pip
  • 3 posts

Posted 17 April 2020 - 02:43 PM

Dear Simon,

 

That statement is very true and with regards to making sure that I am not just fitting my data to the model, I was looking for a scientific backup whether my hypothesis from above is correct.

 

I actually find it rather hard to believe that this issue has not been discussed before, so If you, Simon or any other visitor knows of any literature, I would greatly appreciate any source!

 

Thank you again!

 

Best, Simon






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users