Tactics and Strategies for Managing Ebola Outbreaks and the Salience of Immunization

Figure 1

Our model is a Markov chain branching process in which an individual in state (Exp: exposed/infected but not yet infectious state) can be generated from an individual in state U_Inf (Inf: infectious state) with probability 0 < < < < 1, which is assumed to decrease with time as individuals in the community become more cautious about making casual contact with individuals that have Ebola virus-like symptoms (see SOI Methods for functional forms). Setting the local time of infection of this individual to , this individual becomes infectious at , which we assume to be constant, but can be treated as a random variable with a finite range distribution centered on (e.g., a beta distribution). While infectious on the interval [], this individual may contact and infect other individuals, say one at time —provided this individual is not immune (recovered) or has not been vaccinated with probability increasing over time (see SOI). We assume the infected individuals U_Inf either die or recover and are immune at units of time after being infected (this can also be made a random variable if desired). Here we illustrate several (ignoring Exp or Inf subscript) infected individuals: the index case, the first of the secondary cases, and , an arbitrary general case. Over global time, , we assume that it becomes increasingly likely—with probability 0 < < 1 (see Figure S1 in Supplementary Material available online at http://dx.doi.org/10.1155/2015/736507)—that any individual is isolated from the community while in its Inf state, and it is then able to transmit only to healthcare workers and does so to an arbitrary healthcare worker . The dependence of this probability on , as well as , allows us to consider case detection efficiencies. Additional model assumptions include the following: isolated patients can only transmit to healthcare workers at a rate given by , and infected healthcare workers are isolated immediately on infection.