“…In particular, while Schutz et al's solu-Typeset by REVT E X tion [46] for SIR interactions on a chain with homogeneous initial conditions is a notable exception, typically analytical solutions for any fluctuation-based phenomena associated with SIR systems are absent and more complex numerical approaches are required. For the coarsergrained stochastic differential equation framework, computations typically use Euler-Maruyama [35,47], implicit Euler [38,42], or Milstein [47,48] methods, while for models capable of resolving at the level of the individual one finds numerical techniques based on Gillespie's algorithm [49,50] or direct Monte Carlo simulation [18,22] (see, also, [51]). However, these computational frameworks necessitated by the modelling at the resolution of the individual are not generally amenable to likelihood maximisation or Bayesian inference.…”