5 replies to this topic

[Emily Colby]

Nathan,

 

Take a look at this example. It has sublingual dosing, so there are two dosing compartments (one describing absorption under tongue and the other describing absorption in gut). The compartment for under the tongue is like the immediate release, and it has a fractional bioavailability. The compartment for the gut is like an extended release, it has a lag time, and the bioavailability is the rest of what wasn't absorbed under the tongue. Note that the same amount is applied to both compartments. It was built graphically, but you can switch to text if you like. Let me know if that helps.

 

Best regards,

Emily [file name=PKModel2abs.phxproj size=1767601]http://pharsight.com/extranet/media/kunena/attachments/legacy/files/PKModel2abs.phxproj[/file]

Attached Files

•  PKModel2abs.phxproj   1.69MB   312 downloads

[Emily Colby]

Here is the model text:

 

test(){

    deriv(A1 = - (Cl * C) + (Aa2 * Ka2) + (Aa1 * Ka1))

    urinecpt(A0 = (Cl * C))

    deriv(Aa2 = - (Aa2 * Ka2))

    deriv(Aa1 = - (Aa1 * Ka1))

    C = A1 / V

    dosepoint(Aa2, tlag = (Tlag), bioavail = (1-Frct), idosevar = Aa2Dose, infdosevar = Aa2InfDose, infratevar = Aa2InfRate)

    error(CEps = 0.1)

    observe(CObs = C * (1 + CEps))

    dosepoint(Aa1, bioavail = (Frct), idosevar = Aa1Dose, infdosevar = Aa1InfDose, infratevar = Aa1InfRate)

    stparm(V = (tvV) * exp(nV))

    stparm(Cl = (tvCl) * exp(nCl))

    stparm(Ka2 = (tvKa2))

    stparm(Ka1 = (tvKa1) * exp(nKa1))

    stparm(Tlag = (tvTlag))

    stparm(Frct = (tvFrct))

    fixef(tvV = c(, 20.6, ))

    fixef(tvCl = c(, 1.83, ))

    fixef(tvKa2 = c(, 1.075, ))

    fixef(tvKa1 = c(, 7.62, ))

    fixef(tvTlag = c(, 2.3, ))

    fixef(tvFrct = c(, 0.515, ))

    ranef(diag(nV, nCl, nKa1) = c(0.01, 0.01, 0.01))

}

[Emily Colby]

It looks like the code is similar, so you probably just need to make sure you're applying the amount to both compartments. You may need to duplicate your AMT column so you have two columns to apply.

[Jason Chittenden]

I followed the thread and it looks like you got this model working. My thoughts when looking at it were that it should work. I do see a potential problem with the distribution on F. If you really want F to be bound between zero and one (0,1) write it like this:

 

stparm(F = ilogit(tvF + nF))

 

tvF=0 correspsonds to F=0.5

 

The way that you have it written, F is lognormal and the fat tail on that can cause F>1. If you don't include a random effect on F then you can bound tvF as you did and there won't be a problem.

 

Cheers,

-Jason