Original post on the nlmixr2 project blog is here.
The nlmixr2 Working Group continues to expand what open source R tooling can support in pharmacometrics. In a new post, Matthew Fidler introduces nlmixr2save 0.2.0, now on CRAN. saveFit() writes a finished fit as text you can open later, and the := operator caches a long run so a script or report does not fit the same model again.
A nonlinear mixed-effects fit can take minutes to weeks, and saveRDS() ties the result to R’s binary format and to how nlmixr2 stored the object internally. That tie already broke once, when compressed fits moved off the qs package. saveFit() splits the fit into nlmixr2 and rxode2 source plus CSV tables, bundled in one zip, and loadFit() rebuilds a working fit for diagnostic plots, VPCs, and the next model. While you are developing, changing <- to := hashes the model, the data, and the estimation options, then loads the cache when nothing has changed. The post also covers cache options for a project or a rendered document, and nlmixr2saveShare() for sharing a model when the subject-level data has to stay behind. The same material is on the nlmixr2save poster at ACoP 2026 (October 11–14, National Harbor, Maryland).
Watch the 7-minute narrated walkthrough: https://youtu.be/TD7yhaaQToE
Read the full post from the nlmixr2 blog: nlmixr2save: fits you can read, runs you don’t repeat
Why nlmixr2?
The vision of nlmixr2 is to develop a R-based open source nonlinear mixed-effects modeling software package that can compete with commercial pharmacometric tools and is suitable for regulatory submissions.
In short, the goal of nlmixr2 is to support easy and robust nonlinear mixed effects models in R.
Get involved in the nlmixr2 working group
The nlmixr2 working group is open to anyone interested in contributing to the project. You can get involved by:
Main repo: https://github.com/nlmixr2/nlmixr2
Issues: https://github.com/nlmixr2/nlmixr2/issues
Discussions: https://github.com/nlmixr2/nlmixr2/discussions