Single Model for Influenza Forecasting of Multiple Countries by Multi-task Learning

The accurate forecasting of infectious epidemic diseases such as influenza is a crucial task undertaken by medical institutions. Although numerous flu methods and models based mainly on historical activity data online user-generated contents have been proposed in previous studies, no model targeting multiple countries using two types exists at present. Our paper leverages multi-task learning to tackle the challenge building one countries; each country task. Also, develop prediction with higher performance, we solved issues; finding suitable search queries, which are part contents, how leverage queries efficiently creation. For first issue, propose transfer approaches from English other languages. second novel that takes advantage an attention mechanism extend for countries’ forecasts. Experiments epidemics five demonstrate our significantly improved performance leveraging compared baselines.