I am not saying that this will grow exponentially for ever. For what I know, disease expansion is modeled with a sigmoid curve: exponential at first, then slows down once a saturation level in the population is reached.

According to Wikipedia, the total population of the 3 most affected countries - Liberia, Guinea and Sierra Leone - is roughly 20 millions. So there is at least that upper bound.

The real risk of pandemia comes from the fact that as the number of cases grows, the more likely it is that some of those cases travel through a porous border into other countries which lack the infrastructure and professional discipline to contain new outbreaks. If/when that is the case, each new region will start its own sigmoidal worth.

Well the sigmoid curve you are referring to would be for the case that patients who have contracted the virus and survived would be immune or die. In an SIR model (http://en.wikipedia.org/wiki/Compartmental_models_in_epidemi...) that takes death rate and population number into account you would then expect a disease to either reach a disease free equilibrium or a endemic equilibrium. What I was suggesting is to couple those equations on the country level by modelling travel patterns (flow of Suscebtible Recovered Infected across borders), the result would most likely be that western countries would be able to reach a disease free equilibrium, as they have better recovery rates and lower infection rates, whereas ebola would remain endemic in africa.

Thanks, I was not aware of the SIR model.

Regarding your original question... if you can identify individual regions where the model can be applied, maybe you could use Markov chains to model the transitions between states in each region. I am not qualified enough to do that, but I imagine that the result would be a few endemic nodes where the disease is always present (with different degrees of intensity over time) and from where outbreaks get sporadically "exported" to disease free nodes.