The relationships between message passing, pairwise, Kermack-McKendrick and stochastic SIR epidemic models (Q1679013)

This paper studies the relationships between message passing, pairwise, Kermack-McKendrick and stochastic SIR models. A general stochastic model is proposed which allows all of an individual's post-infection contacts times and the negative of its infectious period to be positively correlated. The message passing system for this model is given. A non-restrictive sufficient condition for the message passing equation to have a unique feasible solution is deduced. It is shown that the message passing system cannot underestimate the expected epidemic size for time \(t>0\). For a special homogeneous stochastic model, the message passing system reduces to a system of four equations. A simple relation for an upper bound on the final epidemic size is proved and sufficient conditions for no major outbreak in this model are established. The classic Kermack-McKendrick epidemic model is derived as an asymptotic special case of the homogeneous message passing system.