Flu Activity Forecasting Website Launched

January 19, 2016 -- CDC today launched a new website that provides forecasts of flu activity made by external research groups. The website is a product of the “Epidemic Prediction Initiative,” a joint effort between federal and external researchers to advance the science of forecasting infectious diseases, including influenza. The website, called “FluSight: Seasonal Influenza Forecasting,” is available at predict.phiresearchlab.org.

CDC currently tracks flu activity through a nationwide domestic influenza surveillance system. This surveillance system informs public health decision-making, but it lags behind real-time flu activity. The goal of infectious disease forecasting is to provide a more-timely and forward-looking tool that predicts rather than monitors flu activity so that health officials can be more proactive in their prevention and mitigation responses. For example, if it were possible to predict when and where flu activity will peak, health officials, health care providers and other partners could plan in advance to optimize the timing of flu vaccination clinics and communications outreach efforts around vaccination, as well as distribution of flu antiviral medications. Similarly, if peak activity could be predicted during a pandemic, health officials could better time the implementation of disease mitigation strategies, such as closing schools, cancelling large social gatherings, or distributing medical countermeasures, such as antiviral medications.

The science of forecasting flu activity is still in its infancy, but CDC continues to support it due to its potential public health value. CDC’s Influenza Division began working in 2013 to advance flu forecasting efforts by engaging with members of the scientific community to develop innovative and cost effective methods to predict flu activity. This effort launched with the “Predict the Influenza Season Challenge,” a contest which encouraged participants to predict the timing, peak, and intensity of the 2013-14 flu season using social media data (e.g., Twitter, internet search data, web surveys, etc.) along with data from CDC’s routine flu surveillance systems. The prize for the most accurate model was $75,000.

CDC used a competition to channel private sector resources into the search for innovative approaches to flu forecasting because this approach was more cost-effective than relying on traditional funding and governmental contract mechanisms. Eleven teams participated in the original CDC competition. Team members developed their own models to predict flu activity based on a variety of data sources. On June 18, 2014, CDC announced the winner of the contest: a team led by Dr. Jeffrey Shaman from the Department of Environmental Health Sciences at the Mailman School of Public Health at Columbia University.

Since the competition, CDC has continued to work with contest participants and a few additional groups that subsequently became involved in flu forecasting activities. During the 2014-2015 influenza season, five of the teams from the original challenge provided forecasts as part of a collaborative, non-monetary challenge. Forecasts by Roni Rosenfeld and his team from Carnegie Mellon University and Jeff Shaman and his team from Columbia University were the most accurate that season and were the winners.

For the 2015-16 flu season, there are currently eight external research groups participating in this initiative. New this season is the FluSight website. Each of the participating teams will provide flu forecasts on a weekly basis throughout the flu season and forecasts will be posted by mid-week. CDC created this central website so that visitors can follow the progress of these predictions and compare them to actual flu activity over the course of the season. Implementation of the FluSight website was funded by the HHS Idea Lab Ignite Accelerator, which supports new governmental projects with resources and other forms of assistance. The website provides data, relevant public health forecasting targets, standard evaluation metrics, forecasts, and best practices. Visitors should note that publication of these models does not signal endorsement by CDC. All of these models will be continually refined and adjusted based on how well the forecasts align with actual flu activity. The various forecasts provided by the website may vary significantly from one another and may be inaccurate.

The launch of the FluSight website coincides with a CDC Public Health Grand Rounds meeting held today that explores more broadly the public health possibilities associated with disease forecasting across a spectrum of different disease threats, including flu, Dengue and Ebola. Presentations from the meeting, which describe the various disease forecasting initiatives underway at CDC, will be made available for online viewing here. CDC will continue to support the development and advancement of flu and other disease forecasting in the hopes that improvements to the science will lead to accurate forecasts that inform public health action in the future.

Related Links:

• CDC Competition Encourages Use of Social Media to Predict Flu - November 25, 2013
• CDC Announces Winner of the ‘Predict the Influenza Season Challenge’ - June 17, 2014