nlmixr

nlmixr is an R package for population pharmacokinetic and pharmacodynamic modeling using nonlinear mixed effects models. It provides a flexible framework to specify structural models, random effects, and residual error models, and to estimate model parameters from longitudinal data collected on multiple individuals.

The package focuses on simulation-based estimation and supports differential equation models through integration with an ODE solver such as RxODE. This enables a wide range of PK/PD structures, including compartmental models and time-varying covariates. Users can implement both differential equation–based and algebraic models and describe inter-individual variability with random effects.

Estimation methods implemented in nlmixr include FOCE and FOCE with interaction (FOCEI), as well as stochastic approximation EM (SAEM). The choice of method depends on model complexity, data size, and the desired balance between speed and accuracy. The toolkit emphasizes fitting population models to data, along with diagnostic and visual assessment tools.

Data input is typically in long format, with identifiers, times, observed concentrations or responses, dosing information, and covariates. nlmixr provides support for various residual error models, covariate effects, and model diagnostics, and it offers plotting capabilities and options for parallel computation to speed up analyses.

nlmixr is open source and available on CRAN and GitHub. It has an active user community with tutorials, vignettes, and example models. The project is widely used for building and evaluating population PK/PD models, performing model comparison, and aiding dose optimization. A newer nlmixr2 revision exists, offering API refinements and performance improvements while maintaining core functionality.