Is it possible to perform the equivalent of MAX=0 in NONMEM using mrgsolve?

[quagmirerules]

I have seen that it is possible to perform some Bayes estimation with mrgsolve. Is it also possible to perform the equivalent of MAXEVAL=0 in NONMEM? If so, how can I do it?

Thanks in advance,

Jason

[dpastoor]

Jason,

Are you trying to get map bayes estimates? Is this what you're after? https://github.com/metrumresearchgroup/mrgsolve/wiki/map_bayes

[quagmirerules]

Hi Devin,

It looks like it is what I am looking for but not sure about the terminology. In NONMEM I can do MAP with foe example: $EST METHOD=IMPMAP INTERACTION. This is not what I am looking for. What I want is simply $EST METHOD=0 MAXEVAL=0. Basically, I want to use a PopPK model with IIV and RV variability to estimate the individual parameters of a particular subject based uniquely in the observed concentrations of the subject. Do you call that map bayes estimates?

In summary:

1) I have a PopPK model.

2) I have plasma concentration from only 1 subject.

3) I want to estimate the individual parameters of this subject based on the observed plasma concentration 2) and the PopPK model 1).

I have also checked there is a maximum likelihood estimation tutorial: https://github.com/metrumresearchgroup/mrgsolve/wiki/simplefit However, it says: "Don't use R + mrgsolve to fit PK models like this when you have an alternative (like NONMEM)." Should I use NONMEM over R+mrgsolve for my purpose? If so, what is the reason?

Thanks,

2016-10-11 14:01 GMT-04:00 Devin Pastoor notifications@github.com:

Jason,

Are you trying to get map bayes estimates? Is this what you're after? https://github.com/metrumresearchgroup/mrgsolve/wiki/map_bayes

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/metrumresearchgroup/mrgsolve/issues/140#issuecomment-252995534, or mute the thread https://github.com/notifications/unsubscribe-auth/AVuJeXSs1u6KRLhW_AyHEzz-Q2AQGCZUks5qy88BgaJpZM4KT682 .

[kylebaron]

Hi @quagmirerules -

Use the link that @dpastoor referenced. Getting the empirical Bayes estimates from a population model works well in R + mrgsolve. Basically, you fix everything and just estimate the ETAs. This is exactly what you are describing.

The note on on the simplefit vignette https://github.com/metrumresearchgroup/mrgsolve/wiki/simplefit

is only to let you know that using R + mrgsolve to fit population models doesn't offer a lot of advantages over just doing it in NONMEM. I have fit population models for client projects and published the results etc. You definitely can do it. But it's probably more hassle than it is worth.

But getting empirical Bayes estimates in R with mrgsolve makes sense: it's fast and flexible and gives you same result as NONMEM.

[quagmirerules]

Thanks a lot for the explanation :)

2016-10-11 14:51 GMT-04:00 Kyle Baron notifications@github.com:

Hi @quagmirerules https://github.com/quagmirerules -

Use the link that @dpastoor https://github.com/dpastoor referenced. Getting the empirical Bayes estimates from a population model works well in R + mrgsolve. Basically, you fix everything and just estimate the ETAs. This is exactly what you are describing.

The note on on the simplefit vignette https://github.com/metrumresearchgroup/mrgsolve/wiki/simplefit

is only to let you know that using R + mrgsolve to fit population models doesn't offer a lot of advantages over just doing it in NONMEM. I have fit population models for client projects and published the results etc. You definitely can do it. But it's probably more hassle than it is worth.

But getting empirical Bayes estimates in R with mrgsolve makes sense: it's fast and flexible and gives you same result as NONMEM.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/metrumresearchgroup/mrgsolve/issues/140#issuecomment-253009648, or mute the thread https://github.com/notifications/unsubscribe-auth/AVuJeeAB2bqFpiEcd8gT6mhj4h6PFR3wks5qy9rAgaJpZM4KT682 .

[kylebaron]

@quagmirerules Np. If you're at ACoP in a couple of weeks, stop by and say hi.