Abstract:
|
The US Centers for Disease Control and Prevention (CDC) has a long history of flu surveillance – measuring where flu is circulating and at what levels. However, prior to 2013, the CDC had minimal experience forecasting flu – predicting where and at what levels flu will be circulating in the future. To gauge flu forecasting capabilities in the US, since 2013 the CDC has hosted an annual flu forecasting challenge called the FluSight challenge. Since its inaugural season, flu forecasting has come a long way – from asking the question, “Can we forecast flu in real-time?” in 2013 to the CDC now putting out weekly communication products to the public based on FluSight forecasts. I have participated in the FluSight challenge each year since 2016, winning the challenge in 2018. In this talk, I’ll provide context about flu surveillance and forecasting in the US, describe what the FluSight challenge entails, chronicle the evolution of the FluSight challenge, discuss the important role the FluSight challenge plays in directing a research agenda, provide high-level modeling and public health takeaways, and speculate about the future of flu forecasting in the US.
|