(S-001) Easy model creation using nlmixr2shiny and nlmixr2lib
Sunday, October 19, 2025
7:00 AM - 5:00 PM MDT
Location: Colorado A
Matthew Fidler – Novartis; Emmanuel Adewuyi – London School of Hygiene and Tropical Medicine; William Denney – Human Predictions; John Harrold – nlmixr2 development team; Richard Hooijmaijers – LAP&P; Anne Keunecke – LAP&P; Theodoros Papathanasiou, – GSK; Rik Schoemaker – Occams; Max Taubert – Novartis; Mirjam Trame – Cetera; Justin Wilkins – Occams
Director Novartis Fort Worth, Texas, United States
Disclosure(s):
Matthew L. Fidler, M.Stat., PhD: No financial relationships to disclose
Objectives: Graphical user interface libraries for non-linear models are useful and found in commercial products like Monolix[1] and Finch Studios[2]. The nlmixr2shiny[3] and nlmixr2lib[4] libraries are developed to allow models to be created from an open-source database of common model structures in nlmixr2[5] format. The nlmixr2shiny[3] library allows a model to be developed using preset model types, and optionally customize the model for specific project purposes.
Methods: The nlmixr2lib[4] library is not only a place to submit, store, and curate nonlinear mixed effects models, but also has functions to create models and add/modify pieces of a model to a prior model structure. The nlmixr2shiny library simplifies model creation with a large variety of models by updating and combining models using model piping instead of enumerating every model variant. With this approach, any naming conventions changes or other changes to the modeling functions are applied to all models in the library. This allows easier maintenance of all models created in nlmixr2shiny[3].
Results: The current implementation of nlmixr2shiny[3] allows 1-3 compartment models to be combined with different types of absorption including iv, first order, Weibull and transit compartment models. The PK models can be simply linked with direct, indirect, and effect compartment models. Once selected, compartment properties, parameter estimates, and statistical models can be added.
Conclusions: The nlmixr2shiny library currently allows easy generation of nonlinear mixed effects models using a graphical user interface. This project is a work in progress, and we intend to allow residual model specification, importing models contributed to nlmixr2lib, and importing models from NONMEM [6] and Monolix [1].
