Influenza pandemics have swept the world numerous times during the last few centuries. Cases of bird flu infecting humans have prompted predictions that we are due for another pandemic soon, but skeptics dismiss such prognostications as panic caused by a misunderstanding of probability. The issue can be reduced mathematically to the question of whether the pandemic process has an increasing, constant, or decreasing hazard function. Historical data on past pandemics can be used to estimate the hazard function using imprecise probabilities, giving upper and lower predictive probabilities of an imminent pandemic, given past waiting times. In order to achieve smoother estimates of the imprecise hazard function, an autocorrelated imprecise Normal prior is proposed.
Keywords. Survival analysis, hazard function, autocorrelated prior.
The paper is availabe in the following formats:
Dept. of Mathematics & Statistics
106 Wiggins Road
2 Saginaw Crescent
Related Web Sites