Time varying parameters

[StevenLShafer]

I want to use RxODE for simulation where the PK parameters abruptly change with events (specifically, the onset/offset of cardipulmonary bypass). I can call RxODE 3 times: before bypass, during bypass, and after bypass. However, there does not seem to be a way of entering a matrix of time-specific PK parameters. Is this possible?

Thank you

[wwang-at-github]

Hi Steve:

Nice to hear from you.

There're multiple ways of doing what you described. One simple way is to use the built-in variable t (for time) and the "if" to denote a change of parameter(s) with time. A snippet of example is given below.

ode = " ... if (t < 10) { CL = CL1; } else { CL = CL2; } ... "

In this case, you'll need both CL1 and CL2 in your parameters.

Hope it helps.

Best, Wenping

[StevenLShafer]

Dear Wenping:

Wow, that was fast!

I understand the code and will give it a try.

The text below translates readily into C. However, can the quoted code include something like “i <- tail(which(pkTimes <= t),1)”. This will return the index to my parameter array. However, I’m not sure how much R code is allowed within the quotation marks. My guess is that R code won’t work, and the easiest approach will be to chain if statements.

Thanks!

Steve

From: wwang-at-github notifications@github.com Sent: Monday, July 29, 2019 7:20 PM To: nlmixrdevelopment/RxODE RxODE@noreply.github.com Cc: Steven L Shafer Steven.Shafer@stanford.edu; Author author@noreply.github.com Subject: Re: [nlmixrdevelopment/RxODE] Time varying parameters (#104)

Hi Steve:

Nice to hear from you.

There're multiple ways of doing what you described. One simple way is to use the built-in variable t (for time) and the "if" to denote a change of parameter(s) with time. A snippet of example is given below.

ode = " ... if (t < 10) { CL = CL1; } else { CL = CL2; } ... "

In this case, you'll need both CL1 and CL2 in your parameters.

Hope it helps.

Best, Wenping

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/nlmixrdevelopment/RxODE/issues/104?email_source=notifications&email_token=AFHWJTYIGHUEDF7HOJMWEJDQB4REVA5CNFSM4IHTAVDKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD3BMXRA#issuecomment-516082628, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AFHWJTZXJ646VXU77E3ABP3QB4REVANCNFSM4IHTAVDA.

[mattfidler]

Hi @StevenLShafer

Just as Wenping suggested there is multiple ways to address this. Another is to add the covariates to the event table.

Using Wenping's to illustrate the alternate approach:

library(RxODE)
mod1 <-RxODE({
    ## central 
    KA=2.94E-01;
    tCL1=1.86E+01;
    tCL2 = 7
    V2=4.02E+01;
    ## peripheral
    Q=1.05E+01;
    V3=2.97E+02;
    CL = cov * tCL1 + (1 - cov) * tCL2
    ## Effects
    Kin=1;
    Kout=1;
    EC50=200;
    ## ODE model
    C2 = centr/V2;
    C3 = peri/V3;
    d/dt(depot) =-KA*depot;
    d/dt(centr) = KA*depot - CL*C2 - Q*C2 + Q*C3;
    d/dt(peri)  =                    Q*C2 - Q*C3;
    d/dt(eff)  = Kin - Kout*(1-C2/(EC50+C2))*eff;
});

## You may use eventTable()
et <- et() %>% 
    et(amt=10000, until=240, ii=12) %>% 
    et(amt=20000, time=120, ii=24, until=240) %>% 
    et(0:240) ## Add sampling

## Now add covariates; You can use any R expressions you want for the
## covariates
et$cov <- 1
et$cov[et$time < 120] <- 0

s <- mod1 %>% solve(et)

s %>% plot(centr, eff)

^{Created on 2019-07-29 by the reprex package (v0.3.0)}

Note

RxODE can take now take NONMEM-style data-frames from R. Therefore your “i <- tail(which(pkTimes <= t),1) should be able to work if you apply it to the data-frame before running the solving routine.

[mattfidler]

For more information you can see the following documentation:

• RxODE events
• RxODE time-varying covariates
• RxODE multi-subject solving
• Speeding up RxODE simulations

[mattfidler]

Let us know if you have any more questions or issues @StevenLShafer