Incidence Prediction for the 2017-2018 Influenza Season in the United States with an Evolution-informed Model

Incidence prediction for the 2017-2018 influenza season in the United States Human infections caused by the seasonal influenza virus impose a large burden on public health in the United States and worldwide. Advanced forecasts of the severity of the upcoming influenza season can contribute to timely preparation for the season, including resource allocation and vaccination campaigns. Existing computational methods have already been developed for this purpose, which can be largely classified into two main classes: the first one focuses on within-season forecasting and relies on updated incidence information as the season develops1; the second generates a prediction for the next season based on the data of the current season2,3. Whereas the former predict absolute severity information and peak timing within season (weeks), the latter produces a forecast with a longer lead time for the next season, but is most accurate for the relative frequency of different lineages (or antigenic clusters) and not for the absolute severity or incidence. Recently, a process-based model (EvoEpiFlu) that incorporates evolutionary information into a modified SIRS system of equations (for Susceptible, Infected, Recovered and Susceptible immune classes in the population) was developed to predict absolute influenza incidence for subtype H3N2 for the United States4. EvoEpiFlu makes use of evolutionary information related to antigenic change based only on sequences to infer H3N2 incidence from the summ...
Source: PLOS Currents Outbreaks - Category: Epidemiology Authors: Source Type: research