nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Does JMP Pro support fitting of nonlinear mixed effects modelling?

Bass_Masri

I have a question from a customer specifically about fitting nonlinear mixed effect models.

Does JMP or JMP Pro have an option for this? Or could you suggest an alternative?

I know that there is Nonlinear Modelling and Fit Curve under Specialized Models but I don't think that accounts for Random and Fixed effect.

Thanks Bass

Accepted answers

All comments

Phil_Kay

Hi Bass,

No, I don't think it will be possible to include random effects in a non-linear model.

Random effects are only supported for ordinary least squares regression, as far as I am aware.

There is an open item in the Wish List for random effects in generalized linear models.

Your customer might want to add a wish for random effects in non-linear models.

How would non-linear modelling with random effects help your customer? What problem are they trying to solve?

Cheers,

Phil

nick_shelton

Hi Bass,

There's a new generalized linear mixed model add-in available: Generalized Linear Mixed Model Add-in

Depending on your customer's use-case it might meet their requirements.

Bass_Masri

Thank you Phil,

The customer is a statistician located in Melbourne Australia and works for the global biotech company CSL who together with their subsidiary Seqirus develop medicines, vaccines and plasma products.

The customer knows the NLMIXED procedure in SAS that can be used to fit these models, but is looking for a solution which life scientists can easily fit nonlinear mixed effects models by using a point and click version of the NLMIXED procedure.

I have reached out to better understand the issue and determine whether there is solution or an alternative.

Thanks again for your advice

Kindest Regards

Bass Masri

Bass_Masri

Thank you Nick,

I have provided the comprehensive link to the Generalized Linear Mixed Model Add-in.

I'll let you know if this meets their requirements.

Kindest Regards

Bass Masri

ian_jmp

Stating the obvious I know, but one might hope that a company like that has access to at least one SAS license. In which case you might think about an add-in to make the SAS Proc more usable by the target group . . .

And I assume that the operative word in this discussion is 'nonlinear', with an emphasis on the 'non'.

Cheers,

Ian

Florian_Vogt

Hi Bass,

I received the same request from the German part of CSL Behring. Maybe my information helps in this discussion.

I also asked for the use case.

1. The do pharmacokinetic studies (PK 1/2-compartment models) with repeated measures. So they have an exponential model want to add random effects to the PK parameters. (I hope I translated well enough)

2. They want to specify their own non-linear model and choose parameters as fixed/random.

@Phil_Kay Do you think this current request is covered by the existing post on the wish list you mentioned? Should I forward this to them for voting?

Best and have a good weekend all

Florian

Phil_Kay

Hi @Florian_Vogt ,

No, I think the wish list item refers only to generalized linear models. Random effects in non-linear models would be a different thing.

Cheers,

Phil