About Flu Forecasting
Influenza (flu) places a significant disease burden on the U.S. population, but the magnitude and timing of flu activity varies from season to season, making the annual impact of flu uncertain at the beginning of each season. Flu forecasting can help decrease that uncertainty by predicting in advance when and where increases in flu activity will occur. Unlike CDC’s traditional influenza surveillance systems, which measure flu activity while it is occurring, flu forecasting offers the possibility of looking into the future and better planning ahead for changes, such as increases in flu-related hospitalizations.
CDC’s efforts with forecasting began in 2013 with the “Predict the Influenza Season Challenge,” a competition that encouraged outside academic and private industry researchers to forecast the 2013–2014 flu season. Since then, CDC’s Influenza Division has collaborated each flu season with external researchers on flu forecasting. CDC has provided forecasting teams data, relevant public health forecasting targets, and forecast accuracy metrics while teams submit their forecasts, which are based on a variety of methods and data sources, each week.
Currently, CDC coordinates forecasts of flu-related hospitalizations each week which are based on data from HHS Protect. Prior to the 2021-2022 season, flu hospitalization forecasts were based on FluSurv-NET data. The change to using HHS Protect data was implemented during the 2021-2022 season because the data from this system can provide a more complete picture of the number of flu hospitalizations in the United States. CDC did not provide flu forecasts during the 2020-2021 flu season because there was too little flu activity to produce stable forecasts.
Flu forecasts can be used to prepare for changes in flu activity, such as increases in hospitalizations. When forecasts accurately predict flu activity, more effective planning of public health responses to seasonal flu epidemics and future flu pandemics is possible. Flu forecasts can inform messaging to health care providers regarding:
- antiviral treatment allocation,
- preparation for an influx of flu-related hospitalizations,
- informing the distribution and placement of health care staff, hospital beds and treatment resources.
Flu forecasts can also be used to help guide personal and community mitigation strategies. These can include non-pharmaceutical interventions, such as reducing contact during times of forecasted high flu activity, as well as conveying the importance of flu vaccination prior to forecasted increases in flu activity.
While significant progress has been made in the years following the initial competition, forecasting respiratory disease remains challenging. Flu viruses are constantly changing, and every flu season tends to be different from the one before it. CDC and participating teams collaborate to identify the best data, methods, and practices for forecasting in order to support the advancement of the science of flu forecasting, improving its ability to inform public health action. These collaborations are critical to developing a network of forecasters providing results that public health officials can use to guide their work. Examples of these collaborations include CDC’s work with Carnegie Mellon University and University at Massachusetts Amherst, which were awarded funding in 2019 to further improve the accuracy and communication of flu forecasts at the national, and state level.
CDC, FluSight partners, and stakeholders also gather at the end of every forecasting season to review forecasting approaches, discuss the accuracy of forecasts from the past season, identity overall challenges and successes and plans for future seasons, such as the additions of new forecasting targets. These meetings improve the usefulness of forecasting by providing the opportunity for collaboration among forecasters and public health officials.