Asymptotic final-size distribution for some chain-binomial processes

From MaRDI portal





This paper generalizes the classical Reed-Frost process of the spread of an infectious disease in a finite population by allowing infection probabilities to depend on current epidemic size. For its analysis the author gives a new representation for a class of chain-binomial processes by sampling a stochastic process X at suitable times. \(X\in D_ R[0,\infty)\) is defined as \(X(t)=\sum^{n}_{i=1}\) \(1_{(Q_ i>t)}\) where \(1_{(\cdot)}\) denotes the indicator function of the event (\(\cdot)\) and where \(\{Q_ i\}_{i=1,...,n}\) are i.i.d. \({\mathbb{Z}}^+\)-valued random variables with d.f. F, n is the number of initially susceptibles and \(Q_ i\) can be interpreted as the i-th susceptible's resistance to infection, assuming that each infective will contribute with one unit to the infection pressure operating on susceptible individuals. The sequence of sampling times \(\{v_ k\}\) is defined as \(v_ 0=0\), \(v_ 1=m\) the number of initially infectives, and \(v_ k=n+m-X(v_{k-1})\), \(k\geq 2\), and represents the cumulated number of infectives up to the k-th generation of infectives. By defining the number of susceptibles \(S_ k=X(v_ k)\) and the number of infectives \(I_ k=v_{k+1}-v_ k\) a chain binomial process \((S_ k,I_ k)\) can be imbedded in X. The final size of the epidemic n-S, where S denotes the number of susceptibles when the epidemic stops, has the same distribution as the time for the first crossing of the linear boundary \(n+m-t\) of the imbedding process X. The asymptotic distribution of this final size is derived from weak convergence results for X. The existence of a distribution determining set of harmonic functions for these chain-binomial processes is also established.




Cited in
(43)








This page was built for publication: Asymptotic final-size distribution for some chain-binomial processes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3705261)