Maximum likelihood estimation in nonlinear mixed effects models

A stochastic approximation version of EM for maximum likelihood estimation of a wide class of nonlinear mixed effects models is proposed. The main advantage of this algorithm is its ability to provide an estimator close to the MLE in very few iterations. The likelihood of the observations as well as the Fisher Information matrix can also be estimated by stochastic approximations. Numerical experiments allow to highlight the very good performances of the proposed method.