Introduction

Designing clinical trials to support population PK/PD modeling requires careful choices of sampling times, number of subjects, dose groups and other trial features to ensure precise parameter estimation – with low relative standard errors [1]. Efficient methods based on the Fisher Information Matrix (FIM) and a first–order approximation around typical population values [2] are implemented in R packages such as popED [3] and PFIM [4]. However, adoption of these existing tools is limited because their model definition language differs from that of widely used NLME estimation software such as Monolix. 

By Géraldine Cellière, Matthias Pierre, Monika Twarogowska, Jonathan Chauvin