NONMEM Users Network Archive

Hosted by Cognigen

Re: [NMusers] Parameter uncertainty

From: William Denney <wdenney_at_humanpredictions.com>
Date: Wed, 15 Feb 2017 07:00:47 -0500

Hi Fanny,

It is often good practice to fit parameters that must be positive on the log=
 scale (by exponentiating them). That will ensure that when sampling from a=
 normal distribution (and then exponentiating the sample) you will have a po=
sitive value.

LLP was suggested, but it won't assess correlation between your parameters w=
hich is often important when running simulations.

Bootstrap is another good alternative as has already been suggested.

Thanks,

Bill

> On Feb 15, 2017, at 5:55 AM, Fanny Gallais <gallais.fanny_at_gmail.com> wrote=
:
>
> Dear NM users,
>
> I would like to perform a simulation (on R) incorporating parameter uncert=
ainty. For now I'm working on a simple PK model. Parameters were estimated w=
ith NONMEM. I'm trying to figure out what is the best way to assess paramete=
r uncertainty. I've read about using the standard errors reported by NONMEM a=
nd assume a normal distribution. The main problem is this can lead to negati=
ve values. Another approach would be a more computational non-parametric met=
hod like bootstrap. Do you know other methods to assess parameter uncertaint=
y?
>
>
> Best regards
>
> F. Gallais
>
>
>
>
>

Received on Wed Feb 15 2017 - 07:00:47 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request_at_iconplc.com. Once subscribed, you may contribute to the discussion by emailing: nmusers_at_globomaxnm.com.