But its a multivariate problem space. Exposure is just one of the variables. Others (mutation rate, types of host, cross-infection rate etc) are also important and highly non-linear.

It's a big-data problem - what are the regression stats for each independent variable? And is 'exposure' the important one?