Dear NMusers,
to get an idea of the uncertainty of my estimated parameters, I recently tried using the llp (log-likelihood profiling) function of PsN (in Pirana). I also created plots using the incorporated R-code. I now wanted to ask if someone knows how I could “force” llp to try further plausible estimate values than just 10 in order to obtain a smoother plot. I have attached the llp plot of my V2 estimate; the curve appears to be very rough and imprecise. Unfortunately, for this specific project I may not use R scripts to do the log likelihood profiling because it specifically aims to compare PsN functions with another algorithm.
Thank you very much in advance for your help.
Kind regards,
Olga Teplytska
Olga Teplytska
- Pharmacist -
Pharmaceutical Institute
- Clinical Pharmacy -
An der Immenburg 4
Room 3.109
D-53121 Bonn
Phone: +49-(0)228-735242
E-Mail: [email protected] / [email protected]
www.klinische-pharmazie.info
PsN_llp_plot_V2.pdf
Description:
Adobe PDF document
llp function in PsN
Dear Olga,
The PsN llp program is aimed at finding the confidence interval with as few
nonmem runs as possible, and in this case the initial guesses were highly
successful.
For this reason you do not get much information on what the curve looks like,
since the first attempt by PsN was close to the target delta-OFV of 3.84 (in
both directions).
If you are really invested in the whole uncertainty distribution, and not only
the confidence limits, then maybe you should consider other available
algorithms?
PsN includes also bootstrap and SIR that would give you the multivariate
uncertainty, and can be used for only looking at uncertainty distribution of V2
if that is what you like?
But maybe it is really shapes of delta-OFV for univariate parameter value that
you want to investigate further?
I cannot suggest a way to investigate in PsN a whole grid of parameter values
without writing a few lines of code, or preparing a whole set of control
streams with different fixed values for the parameter in question, but maybe
someone else can suggest a PsN command for this?
Best regards
Jakob
Jakob Ribbing, Ph.D.
Principal Consultant, Pharmetheus AB
Quoted reply history
> On 7 Feb 2023, at 10:31, Olga Teplytska <[email protected]> wrote:
>
> Dear NMusers,
>
>
> to get an idea of the uncertainty of my estimated parameters, I recently
> tried using the llp (log-likelihood profiling) function of PsN (in Pirana).
> I also created plots using the incorporated R-code. I now wanted to ask if
> someone knows how I could “force” llp to try further plausible estimate
> values than just 10 in order to obtain a smoother plot.
> I have attached the llp plot of my V2 estimate; the curve appears to be very
> rough and imprecise.
> Unfortunately, for this specific project I may not use R scripts to do the
> log likelihood profiling because it specifically aims to compare PsN
> functions with another algorithm.
>
> Thank you very much in advance for your help.
>
> Kind regards,
> Olga Teplytska
>
>
>
> Olga Teplytska
> - Pharmacist -
>
> Pharmaceutical Institute
> - Clinical Pharmacy -
> An der Immenburg 4
> Room 3.109
> D-53121 Bonn
>
> Phone: +49-(0)228-735242
> E-Mail: [email protected] / [email protected]
> www.klinische-pharmazie.info
>
>
> <PsN_llp_plot_V2.pdf>
--
*This communication is confidential and is only intended for the use of the
individual or entity to which it is directed. It may contain information
that is privileged and exempt from disclosure under applicable law. If you
are not the intended recipient please notify us immediately. Please do not
copy it or disclose its contents to any other person.*
*Any personal data
will be processed in accordance with Pharmetheus' privacy notice, available
here https://pharmetheus.com/privacy-policy/.**
*