Phoenix NLME and Phoenix
Dan Weiner and Simon Davis
Pharsight Corporation
Objective: To demonstrate
- Phoenix NLME (beta) for nonlinear mixed effects modeling
- Phoenix Connect for supporting usage of NONMEM, SAS, S-Plus and SigmaPlot in Phoenix workflows
Results: Phoenix NLME is capable of fitting a wide variety of non-linear mixed effects models and datasets. Estimation methods include FO, FOCE (ELS and Lindstrom-Bates), Laplacian, adaptive Gaussian quadrature, naïve pooled and iterative two-stage as well as a sophisticated non-parametric algorithm. Support for covariate selection, model selection and model qualification (bootstrap and posterior predictive checks) is also provided. Results are generally comparable with NONMEM, but NLME often executes faster as its engines are parallelized. Results can be compared to NONMEM side-by side using Phoenix Connect.
Conclusions: Phoenix NLME (to be released later this year) and Phoenix Connect provide a powerful modeling and simulation environment for nonlinear mixed effects problems. Use of workflows facilitates efficiency via creation of report ready graphs and tables, and can be utilized on multiple datasets.