Exact Gradients Improve Parameter Estimation in Nonlinear Mixed Effects Models with Stochastic Dynamics

H. K. Ólafsdóttir, J. Leander, J. Almquist, M. Jirstrand. In proceedings of The 8th American Conference of Pharmacometrics (ACoP8), October 2017, Fort Lauderdale, USA.


Nonlinear mixed effects (NLME) models based on stochastic differential equations (SDEs) have evolved into a mature approach for analysis of PKPD data [1-3], but parameter estimation remains challenging. We present an exact-gradient version of the first order conditional estimation (FOCE) method for SDE-NLME models, and investigate whether it enables faster estimation and better gradient precision/accuracy compared to finite difference gradients.



Photo credits: Nic McPhee