What Makes Pandemic Prediction Difficult?

Briefly, the primary objective of this project was to identify methods that disrupt and/or enhance the predictability of infectious disease dynamics (IDD) of emerging outbreaks. We specifically focused on the variation in final sizes (total number of infected individuals) due to differences in contact networks, with plans to extend our framework as incidence over time versus focusing solely on the epidemic final siz]. Our focus on contact networks is based on the assumption that transmission parameters can be readily measured early in an epidemic. As a result, while contact patterns are difficult to measure, they are the pragmatic force driving uncertainty in the predictions of outbreak sizes. We also were interested in using a variation of a GAN approach to train a “prophet” CNN (connected to a fully connected FFNN) prediction model that “fought” with a VAE demon model.