17 replies to this topic

[Georgia Charkoftaki]

Hello,

I have developed a joint parent-metabolite model for cyclophosphamide. I've run separate POPPK models for the parent and the metabolite (4-OH Cy) which were robust. I found a correlation between Kel (parent drug) and Kform (metabolite, but I use Ka symbol in my code), so I added a linear correlation. My issue is that I keep getting a point in my graph (-5.33, -5.33), which I can't find in my data and I don't know whether I set up my data sheet correctly, since I 've added BQL and MDV model #2. I also don't know if because of this point I get high sd.

Could you please help me by taking a look in my project? I would really appreciate your feedback.

Thanks,

Georgia

 

P.S. I am trying to upload my project here and I can't.

[Samer Mouksassi]

hi make sure that you select file to attach instead of select image to attach

Samer

[Georgia Charkoftaki]

I know- I have been trying forever and no luck!

[Georgia Charkoftaki]

I think the file was too big. [file name=Metabolite_parent.phxproj size=8338717]http://www.pharsight.com/extranet/media/kunena/attachments/legacy/files/Metabolite_parent.phxproj[/file]

Attached Files

•  Metabolite_parent.phxproj   7.95MB   345 downloads

Edited by Simon Davis, 08 September 2015 - 08:46 AM.

[Samer Mouksassi]

I see you included the time zero and Concentration = 0 point.

We cannot measure a concentration of exactly zero it is probably BQL why the BQL flag is not set to one for these observations ?

[serge guzy]

Dear Georgia

There is also an error (in addition of Samer's comment) also in your code

 

fixef(dkadCL) = c(, 0.5, ))

 

should be

fixef(dkadCL = c(, 0.5, ))

 

I am not sure it is related but the code in your project is definitively wrong.

 

Strange it did not crash.

Best

Serge

[serge guzy]

Another code error

 

ranef(diag(nVmet, nCLmet, nKa, dkadCL) = c(0.04,0.04, 0.04, 0.04))

 

should be

ranef(diag(nVmet, nCLmet, nKa) = c(0.04,0.04, 0.04))

 

 

dkacl is a fixed effect not a random effect and you defined it through fixef statement.

best

Serge

[serge guzy]

Use foce els not foce lb.

Best

Serge

[serge guzy]

Can you try the following initial estimates

 

tvVmet 4956390

tvCLmet 250934

tvKa 0.0621264

dkadCL 1.03951

stdev0 0.385595

 

 

Best

Serge

[serge guzy]

When you have BQL you need to put 1 as you did in the column CObsBQl but the LOQ value in the response column.You put BQL which is not recognized by NLME.

BEST

SERGE

[serge guzy]

Not sure you got that but BQL in your data set in the response column should have the LOQ value (limit of quantification) and 1 (you did that) in CObsBQL column.

Best

Serge

[serge guzy]

Dear Georgia

Please look at the site

https://app.box.com/s/fv5vvi7d79xzwz0x69by

 

I have there a slightly different version of your model. Since you did not put the LOQ value I did the model without BQL. Once you have the LOQ value put it in the CObs column whenever CObs BQL is 1. Note that in your version once you have BQL you will not get the CWRES plots.

Best

Serge

[serge guzy]

https://app.box.com/s/fv5vvi7d79xzwz0x69by

 

That is the location

[Georgia Charkoftaki]

I have the LOQ value of the method is 0.0 ug/mL, but I didn't know exactly how to incorporate that in the work sheet.

[Georgia Charkoftaki]

Also, don't take into account the model with the covariates, model #2 is the one I was trying to run first and then I would add the covariates.

Thanks,

Georgia

[Georgia Charkoftaki]

Serge,

I use for deriv(A1 = - (Ka* A1)

 

fixed individual values of CL/V, which I have taken from the parent drug and I am trying to correlate with the metabolite.

[serge guzy]

Dear Georgia

Can you give me the exact name of the model you are trying to use?

I see 7 different models. Anyway I think that both Samer and my advises should enable you to successful analyze your data. Let us know if you still have trouble.

Best

Serge

[Georgia Charkoftaki]

Dear Serge,

The joint model is #2. Thank you all for you comments and the support. I will try to improve my data sheet and make that model work.

 

Georgia