Re: [NMusers] RE: Estimation of ke0 from raw data

Hi Jos, Leonid has a post on it from some time ago, where he shows code to do indeed linear interpolation. see http://www.cognigencorp.com/nonmem/current/2008-January/4167.html See also some older thread that discusses more possibilities: http://www.cognigencorp.com/nonmem/nm/99sep241996.html An

Re: [NMusers] eigenvalues

Hi Pavel, Principal component analysis can be validly performed on any matrix, and it is just a matter of convention that the eigenvalue ratios of min/max of the total covariance matrix of estimation are reported as the condition number for a given model. This as a metric of how easily the dimensi

Re: [NMusers] Problem of STS in NONMEM

Dear Anyue, It has been a long time since I used this method. Nowadays NONMEM has much added functionality, such as the non-parameteric step and the possibility to repeat problems over individuals in your dataset. My advise would actually be to use those. The latter especially gives you a much cle

Re: [NMusers] Mixture model with logistic regression

Hi Mark, Is it indeed a logistical model or is it an ordered categorical? I assume you refer to the latter. Not sure how you get your second category otherwise. Anyway, to me it reads like you are trying to have the mixture model describe exactly what the omega is trying to describe. Perhaps

Re: [NMusers] EM methods giving different outcomes at each repeated run

Dear Edgar, What you describe is high stochastic noise in the EM algorithms. The sampling integration needs to be accurate enough in order for these algorithms to work properly. The first parameter to tune is ISAMPLE of SAEM, I would start with 10 if you have less than 20 subjects. Accuracy i

Re: [NMusers] Different results with ADVAN4 and ADVAN6

Dear Hanna, You might have hit a situation where the solver routine (ADVAN6) does not provide an adequate solution to your $DES system (even though a better fit). My advice would be to decrease the error tolerance of the solution by increasing the number that now says TOL=5 until (up to 12 o

[NMusers] Open pharmacometrics position at PD-value

Dear all, PD-value, based in de Uithof (Utrecht Science Park), Netherlands, is looking for an enthusiastic researcher to join its growing team. *What do we do at PD-value?** *PD-value is involved in translational medicine and general pharmacometrics. We are active across all phases of drug re

Re: [NMusers] Easier IOV modeling

Hi Sebastien, Interesting notion, thanks. It should be possible from NM7.3 onwards using MOD or DECLARE as described in the sections just below the one you referenced. With DECLARE it could look like this: $ABBR DECLARE INTEGER IOCC $PK CALLFL=-2 IF (NEWIND < 2) IOCC=0 IF (AMT > 0 .and. CMT=

Re: [NMusers] NonMem and EVID=2

Hi Johan, If you could share your code and a relevant snippet of your dataset you might get more specific help. Without such detail: Bill's suggestion (time-varying covariate) could extend to the covariance step as well - those calculations might be more sensitive to the different values gene

Re: [NMusers] AMD vs Intel

Hi Leonid, "A while" back we compared model development trajectories and results between two computational platforms, Itanium and Xeon, see https://www.page-meeting.org/?abstract=1188. The results roughly were: 1/3 equal, 1/3 rounding differences and 1/3 real different results. From discussio

Re: [NMusers] M3 method - WRES, and CWRES

Hi Mutaz, Bill, It might be useful to use NPDEs, as discussed in https://www.cognigen.com/nmusers/2019-February/7376.html;  the whole thread is worthwhile reading. NPDEs can be calculated also for BQL values. Bill -thanks for pointing to excellent post of Matt! I would take as most important

Re: [NMusers] SEED status with SAEM method

Dear Le Ba, The most optimal way would be to tune other SAEM settings to see whether the results become more insensitive to the seed. Your use of a lower NBURN might be because you do not reach convergence. If that is indeed the case, that would also an objective to reach by improving settings

[NMusers] Open pharmacometrics position at PD-value

Dear all, PD-value, based in de Uithof (Utrecht Science Park), Netherlands, is looking for an enthusiastic researcher to join its growing team. Would you like to seize this opportunity? *What do we do at PD-value?** *PD-value is involved in translational medicine, general pharmacometrics and

Re: [NMusers] OFV by endpoint of joint models?

Hi Matts, The easiest way to assess is when one of two endpoints is modeled directly (TTE, logistic regression) as often is the case, than look at the Y value for those endpoints, as reported in the PRED variable. The sum of those values is the ofv, or proportional to it, for that particular

[NMusers] Open statistical modeler position at PD-value

Dear all, PD-value, based in the Utrecht Science Park (Uithof), Netherlands, is looking for an enthusiastic researcher to join its growing team. Would you like to seize this opportunity? ** *What do we do at PD-value? *PD-value is involved in translational medicine, general pharmacometrics

Re: [NMusers] Condition number

Dear Ayyappa, A nice discussion! It may be worthwhile to inspect further collinearity statistics, see e.g. https://cran.r-project.org/web/packages/olsrr/vignettes/regression_diagnostics.html , with for example VIF and CI being sometimes useful to detect problematic parameters in my experienc

[NMusers] Open pharmacometrics position at PD-value

Dear all, PD-value, based in the Utrecht Science Park (Uithof), Netherlands, is looking for an enthusiastic researcher to join its growing team. Would you like to seize this opportunity? *What do we do at PD-value? *PD-value is involved in translational medicine, general pharmacometrics and

Re: [NMusers] BQL simulations

Dear Hiba, What is your complete error model? The information you provide is too limited to be sure how to help you. However, assuming you used a proportional error model with an additive term for M3/BQL, you might get better results with an exponential error model. For this, you need to u

Re: [NMusers] Observed (yaxis) vs Predicted (xaxis) Diagnostic Plot - Scientific basis.

Hi Joga, Wilbert, It indeed is an interesting aspect. I was triggered to think about this during my masters research (with dense time series), and for me it was helpful to think about orthogonal regression. One can find and compare the expressions in the wikipedia entries at https://en.wikipe

Re: Fwd: [NMusers] Observed (yaxis) vs Predicted (xaxis) Diagnostic Plot - Scientific basis.

Hi James, To answer your question re censoring in the VPC: not necessarily - you can apply the censoring also in the simulation / processing code explicitly. With PsN, this is easily accessed for BQL values with the -lloq/-uloq options, see the section on BQL data in the psn userguide. This a

Re: [NMusers] Verbatim Code

Nowadays you can just specify CWRES in the $TABLE statement of NONMEM to get those residuals, circumventing the issue you ran into. See the nonmem75 pdf for further residuals that are available such as NPDE. (if you really want to investigate what goes wrong, go into the NM_run1 folder and look

Re: [NMusers] NONMEM Error on SAEM but not FOCEI

Dear Omar, What you describe is not that strange. It can happen because SAEM does random sampling whereas FOCE-I uses a directed search. Therefore extreme values can occur with SAEM. Please refer to the section I60 of user guide for v7+ "Stable Routines for Estimation Methods and Automated 

Re: [NMusers] NONMEM Error on SAEM but not FOCEI

wrote: This part is easy to solve: if you do it in log space, Q will always be positive, no PROTECT is needed (and I am not sure it is relevant in the case when Q is going to be negative). On 1/26/2024 10:27 AM, Jeroen Elassaiss-Schaap (PD-value B.V.) wrote: Dear Omar, What you describe is not

Re: [NMusers] NONMEM related implementation question

Hi Santosh, The prediction (PRED/IPRED) is supplied by the $TABLE step for all records, whether EVID=0 or EVID=2, so I do not understand your question as such. Perhaps you can share model code and a dataset stub for us to help you better? Jeroen http://pd-value.com jer...@pd-value.com @PD_v

Re: [NMusers] Warning obtained while running SAEM

Dear Tim, Have you tried to run model 1 (with SQRT) also using $ABBREV PROTECT? The error may occur because of a numeric error that propagates through the SAEM steps. Jeroen http://pd-value.com jer...@pd-value.com @PD_value +31 6 23118438 -- More value out of your data! On 05-03-2024 14:08,

Re: [NMusers] %RSE for IIV when expressed as %CV instead of variance

Dear Anita, You can find all details in {shameless plug} our PSP tutorial on log-normal variability: https://ascpt.onlinelibrary.wiley.com/doi/full/10.1002/psp4.12507 or go straight to the cheatsheet https://ascpt.onlinelibrary.wiley.com/action/downloadSupplement?doi=10.1002%2Fpsp4.12507&fil

Re: [NMusers] %RSE for IIV when expressed as %CV instead of variance

Hi Nick, Thanks for bringing this up. Stuart Beal rightly pointed out that with larger variability the interpretation of the lognormal becomes more involved and merits additional discussion. Nowadays we also seem to encounter values of OMEGA much larger than those that he seemed to consider (

[NMusers] Call: beta testing IMPRES-M

Dear nmusers, We are thrilled to introduce IMPRES-M®, a novel statistical method for PK-PD modeling that employs smoothing. Some smoothing approaches have been published in our field but have not been very beneficial. IMPRES-M®, however, directly yields a smooth curve representing the RESpons

Re: [NMusers] Obtaining RSE%

Dear NMusers, This is a great reminder for us to consider the reliability of standard errors in our models, thanks Ken & Alan. The more non-linear the models become, the less reliable and the more important other perspectives on parameter values such as sensitivity analysis and prior knowledge

Re: [NMusers] Separation of the PK parameters for mixed endogenous and exogenous substances.

Dear Karem, The short answer is that it is possible to set up an useful PK model in this circumstance. The focus however will be to describe the endogenous substance with sufficient detail and quality. Only once you have done that properly you can derive a meaningful PK model for the exogenous

Re: [NMusers] $TABLE issue when using parallel computing and SAEM/IMP

Hi Dongfen,I cannot remember having problems specifically with the table step. Plenty runs were successful with SAEM/IMP in multi core. Have you tested the control stream without table step to confirm it to be the problem?It would be better to see the whole control stream. But perhaps… do you have

Re: [NMusers] RE: Historical question

Hi Douglas, On top of this, it might be the encryption only started with version 7. I remember looking at the FOCE code myself, the earliest version I installed was 5 so that one was open. I believe 6 was open as well, perhaps someone can confirm that. Good luck with your paper! Jeroen htt

[NMusers] Open PK-PD/pharmacometrics position at PD-value

Dear all, PD-value, based in the Utrecht Science Park (Uithof), Netherlands, is looking for an enthusiastic researcher to join its growing team. Would you like to seize this opportunity? *What do we do at PD-value? *PD-value is a dedicated modeling and simulations company works in partnershi

Re: [NMusers] NONMEM error

Hi Dennis,You could try any of the following to prevent the suspected numerical issue:- insert $ABBR PROTECT- set ONLY OBSERVATIONS in the error block - insert a reset event (EVID=4) after the 30-day hiatusHope this helps,Jeroenhttp://pd-value.comjer...@pd-value.com@PD_value+31 6 23118438-- More va

[NMusers] PD-value's hands-on workshop on IMPRES-M at PAGE2025

Dear nmusers, Are you looking to enhance your PK-PD modeling workflows with a faster and more flexible approach? At PAGE 2025, we invite you to our workshop on IMPRES-M, a novel method that leverages smoothing techniques with P-splines to efficiently describe PK and PD curves—allowing for sim

Re: [NMusers] All PRED=0 after removing IIV

Hi Dongfen, I agree with Leonid, and it is likely that $ABBREV PROTECT will do the job. I however would encourage you to make an anonymous version of your model so you can share it with the group if it does not solve the problem. It might be informative to run the table step without estimatio