EMILY COLBY provided the Phoenix model to me for the Savic Transit model with a lag time: it works with transit compartments and it provided an excellent fit to single dose data. The model does not display graphically but the code is there in text. Most probably it was in NONMEM originally and converted over.
I decided I wanted to see what the multiple dose data looked. I cannot see how to get multiple doses in to the system. I am thinking that you can get them without altering the text code in other words use the Phoenix framework to introduce the doses in rather than altering the code.
Advice on steps is most welcome. I show the code from Emily below,
ANGUS
SAVIC CODE
test(){
note we dose into the absorption compartment -
It does not matter whether this is first or last in the sequence of transit and absorption cmts
covariate(AMT)
Here R.Savic numbered the transit compartments as 0, 1, 2, …, Ntran
so there are actually Ntran+1 total transit compartments, with
the amount in the last transit compartment as a function of time=t given
by AMT*(1/Ntran!) exp(-Ktrt)(Ktr*t)^Ntran
where AMT = initial dose at time t=0 into transit compartment 0.
#Note exp(-lgamm(Ntran+1) ) = 1/Ntran!; #amtintrans is AMT*(1/Ntran!)(Ktrt)^Ntran = amount in last transit compartment Ntran as a function of time=t
amtintrans=AMTexp(-lgamm(Ntran+1)+Ntranlog(Ktrt)-Ktrt)
I want to update my recent post to include more information: I have TWO LAGTIME MODELS in one Phoenix file.
Model (a) was one I wrote originally in WinNonlin code as Diff Equations with a number of lag times e.g. 6 as needed. It worked well. When Phoenix models emerged Emily Colby converted it over to Phoenix for me using the graphical approach. It works very well and you can simulate multiple doses according to the usual procedure in PHOENIX.
Model (2) is the Savic model and Emily included that option in the file: it also works well. It was supplied as code and I think the code may have been converted over from NONMEM.
The dose is referred to in the code as AMT and has been added as a column next to the time and plasma concentration columns at zero time. I have tried to simulate multiple doses and get the program to run but I have failed. I am looking for suggestions to get this model to run with multiple doses.
I attach the model as a Phoenix file; suggestions are most welcome.
No I do not have the NONMEM code, but have a presentation from Savic; he is in Sweden. I am speculating that Emily had the NONMEM code from Savic and converted it over to pml, but I do not know if that was the case. I am also thinking as it is written it is for a single dose. I am thinking if it works for a single dose then it can be made to work for a multiple dose application. I developed my multiple lag time model in WinNonlin~ 7 years ago; it was inspired by Peter Bonate’s s book of simulation and modeling.
Savic offers to supply the code on request. Should I ask him for it?
Hi ANgus, thanks for the Savic paper, I have a couple of other things to get finished this week so I haven’t had a chacne to fully digest this however it seems the original model was indeed coded with an explicit analytic solution to the transit model.
The equations with fittable, possibly fractional, number of compartments , and the corresponding Phoenix translation as coded by Emily, do not allow for multiple doses, however re-writing the model using the transit statement means it could handle them. An illustration is Mike Dunlavey’s Phoenix transit model presented as a PAGE poster in 2013.
The fragments below illustrate a first-order model with time-lag and a corresponding transit compartment model:
Please take a look at this poster and the transit statement in the user guides and see how you get on, if I get my other stuff wrapped up I’ll try and come back to this on Friday afternoon.
The Savic equation to me cannot apply to multiple doses the way it is written because it assumes as initial conditions amount given in the transit 0 compt (AMT) and all other compts are empty. Once you add another dose, then you will have accumulation in each of the transit compartments and the equation is not valid even if you would adjust it by defining t-TAD rather than time (Time since last dose). The transit compartment statement fits perfectly for your needs. If you still need clarification, let me know and I will prepare for you an example that uses the transit statement. note that it applies to phoenix 1.4 only, not phoenix 1.3.
one more thing. If we assume linear kinetic, then we may use that equation with t-time since first dose instead of time and use the principle of superposition and accumulate all the concentrations associated with each dose. But I still believe the easiest way is juts to use the transit statement.
