Forecasting the 2013–2014 Influenza Season Using Wikipedia
Fig 4
Seasonal SνEIR fit to 2006–2007 U.S. ILI data.
Ten seasonal heterogeneous SνEIR model parameterizations for the U.S. ILI 2006–2007 data. These are approximate solutions to (9). For each of the influenza seasons, from 2003–2004 through 2012–2013, fits similar to the above were generated. These parameterizations formed the basis for our prior, π0(p). This is a good example of the seasonal SνEIR model’s two areas of systematic divergence. In the weeks 50–1 there is a first peak that the model does not catch. However, the fitted model does envelop the secondary peak around the 8th epidemiological week. During the tail weeks 15–20 our SνEIR model tapers too quickly.