Is volume capacity required to fit PK data in simbiology?

2 views (last 30 days)
I have several time-concentration data obtained from rat plasma and wish to fit a standard 2-compartment model using simbiology. Do I need to specify the compartment capcacity (peripheral and central)? or can I let the default values of 1 liter (which of course do not reflect the actual compartment volumes in rats).
In case a good fit requires the capacities be included, where I could get such data?

Answers (2)

Arthur Goldsipe
Arthur Goldsipe on 9 Nov 2021
Let me first make sure I understand your question. I think you want to estimate the volume of peripheral and central compartment volumes in a 2-compartment model (presumably along with other kinetic parameters). And you want to know whether you can use initial estimates of 1 liter, or whether you need to specify "better" initial estimates.
If that's the case, then in general it is best to use reasonable (that is, physiologically realistic) values for your initial estimates. Depending on your data and the specific optimization method you pick, you may still be able to successfully estimate parameters when your initial estimates are far off. But even is such cases, the optimization will typically take longer than if you start with reasonable estimates.
I think you should be able to get some idea of physiological relevant volumes by searching the literature or even just a general internet search. (I quickly found a couple of rate volumes listed in the 0.05 to 0.15 milliliter range.)
In this particular case, you also only need to estimate a few parameters, so you may be able to determine reasonable initial parameter values my manually exploring the effect of parameter values on simulation results. You can do this easily in the SimBiology Analysis App by adding sliders. This approach is illustrated in a tutorial that you might find useful (at around 2:40 time mark).
  1 Comment
Fulden Buyukozturk
Fulden Buyukozturk on 10 Nov 2021
In addition to what Arthur already suggested, you could also perform NCA on the data and consider using NCA results as initial estimates.

Sign in to comment.


Jordi Alzina
Jordi Alzina on 10 Nov 2021
Edited: Jordi Alzina on 10 Nov 2021
Thanks Arthur and Fulden, your answer helps.
As a matter of fact, I had seen the tutorial you refer to before posting my question. In that particular example (one-compartment model with first order elimination) the NCA delivers good estimates, probably due to the simplicity of the model.
In my particular case I have a very low kel (during the terminal phase) as compared to a very high kel (during the distribution phase). Accordingly the NCA delivers very large compartment volumes, which are not physiological. A model with at least two compartments seems to be a good approach to model the data.
My question is whether in symbiology, the compartment capacities do reflect actual physiological values; or contrary to that, reflect the capacities that one typically obtains when doing PK analyses, e.g. very large peripheral capacities in case liposoluble drugs with slow elimination due to slow transfer from adipose tissues into main circulation.
See below a plot showing the dataset I’m dealing with.
Jordi
  1 Comment
Arthur Goldsipe
Arthur Goldsipe on 10 Nov 2021
I guess I would answer your question by saying that the parameters you estimate in SimBiology will only be as good as your model and data. If your model and estimated parameters fit the data well but the parameters are not physiological, then there could be several explanations.
One explanation is that your model could be physiologically inappropriate for your particular application. Another exaplanation is that your data might not be sufficient for precisely estimating the parameters. For example, in multicompartment models it can be difficult to estimate all parameters when you only have concentration measurements for one of the compartments.
I also sometimes see "non-physiological values" interepreted as "effective values." But if one of your goals is to estimate physiological volumes, it's possible that you will need richer data and a more complex model.

Sign in to comment.

Communities

More Answers in the  SimBiology Community

Products

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!