2 replies to this topic

[csheme]

I have a question on the evaluation of my inhibition of production model and parameters.

 

I have a 2-cmpt model with and without effect compartment added, indeed the IC50 estimates for both are quite different but the CV for both is very tight. Will the IC50 for the model with effect compartment added be representing the central compartment, or will it be the IC50 of the effect compartment. How do we give physiological meaning or translate this to an actual concentration.

 

When reporting typical PD parameters for these models, is k1e, IC50, and kout reported. What is the role of k1e (rate to the productive compartment), and (kin and kout) on the inhibitory response?

 

What about PD t1/2 , can we deduce PD t1/2 using an effect compartment added into an indirect response model?

 

In the model for the indirect response without effect, we see IC50 seems to be of the central compartment, while after adding the effect compartment, the IC50 depends on Ce used like the biophase distribution. 

 

Another question I have is in regards to setting baseline as a covariate on kin, of course baseline=kin/kout so we do have a strong correlation, when i do enable this covariate on PD I see the PD parameter estimates do not change, but the Log likelihood is significant improved (90points), and the CV and BSE are improved for the PD parameter estimates. Is this common practice to allow baseline on kin as a covariate? Basically reducing any correlation in continuous covariate plots for PD parameters. I have left this out of the model for simplicity, but definitely want to have your opinion.

 

I am attaching my model within and want to ask your input if you think the effect compartment addition is justified and if the model structure (built with graphical editor) looks right to you. By my observations the addition significantly improves the log likelihood. In my attached model, under the last workflow there are 2 models (one without effect compartment and one with), could you please have a look at both and provide some input about the model I created.

 

Thank you so much for your advice!

Edited by csheme, 31 May 2016 - 03:09 PM.

[Simon Davis]

Hi Colby, I just wanted to close the loop on this discussion; Serge had looked at your project and suggested a new model but hadn't followed your data transforms fully. I attach a copy of this project with the last model, I think, correctly mapped. Simon

Edited by Simon Davis, 31 May 2016 - 03:17 PM.

[csheme]

Hi Simon, 

 

The model I have seems to fit the data much better then Serge's, his LL and diagnostics are further off (~150). 

Edited by csheme, 31 May 2016 - 09:50 PM.