Deviation metric for overall fit in forecasts of several past influenza seasons.

<p>Overall fit was calculated under dynamic (blue) and static (red) forecasts assuming a specific infectiousness profile function and scalar of 0.50 for pre-existing immunity. The metric was calculated as the % error between observed and simulated epidemics with 95% confidence intervals. For observed data we used the actual number of laboratory-confirmed samples (y-axis). In both types of forecasting, the simulated epidemic curve (scaled) was matched to the first of two consecutive weeks, in the observed epidemic, when laboratory surveillance reported 5 or more positive viral culture samples. Given our definition of the epidemic start week, this corresponded to index week 6 since the start week of the epidemic.