Likelihood Estimation for Stochastic Differential Equations with Mixed Effects

Stochastic differential equations provide a powerful tool for modelling dynamic phenomena affected by random noise. When time series are observed for several experimental units, it is often the case that some of the parameters vary between the individual experimental units. This has motivated a considerable interest in stochastic differential equation models with mixed effects, where a subset of the parameters are random, because these models enable simultaneous representation of randomness in the dynamics and variability between experimental units. When the data are observations at discrete time points, the likelihood function is only rarely explicitly available, so for likelihood‐based inference to be feasible, numerical methods are needed. We present Gibbs samplers and stochastic EM algorithms based on augmented data obtained by the simple method for simulation of diffusion bridges of Bladt and Sørensen (2014). This method is easy to implement and has no tuning parameters. The method is, moreover, computationally efficient at low sampling frequencies because the computing time increases linearly with the time between observations. The Gibbs sampler as well as the EM algorithm are shown to simplify considerably for exponential families of diffusion processes, including a large part of the models that are used in practice. In a simulation study, the estimation methods are shown to work well for Ornstein–Uhlenbeck processes and t‐diffusions with mixed effects. Finally, our methodology is applied to neuronal data, and it is outlined how the general algorithms can be extended to models with measurement errors.