- Predictive analytics models were able to forecast trends in influenza outbreaks with greater accuracy than historical baseline models.
Led by biostatistician Nicholas Reich, researchers at the University of Massachusetts Amherst formed a group called the FluSight Network. The team compared the forecast accuracy of 20 predictive models to a historical baseline seasonal average, using data from influenza seasons in 2010 through 2017.
The team found that the predictive analytics models achieved greater accuracy than other approaches.
“Across all regions of the United States, over half of the models showed consistently better performance than the historical baseline when forecasting incidence of influenza-like illness one, two and three weeks ahead of available data and when forecasting the timing and magnitude of the seasonal peak,” the researchers said.
The group noted that influenza affects approximately nine million to 35 million individuals in the US each year and is a contributing cause to between 12,000 and 56,000 deaths annually. So far this flu season, more than six million have already caught the illness, and as many as 80,000 have ended up in the hospital, Reich added.
The rise of big data has fueled interest in the idea that advanced analytics could lead to improvements in how disease transmission is measured, forecasted, and controlled, the researchers said.
In addition to the comparison of the models, the FluSight Network facilitated the development of an “ensemble model,” which includes all of the network’s forecasted flu trajectories for the year. These individual forecasts are combined into a single ensemble forecast that is sent to the CDC each week.
The CDC and other public health officials use these forecasts to effectively plan and respond to the evolving flu epidemic. The CDC also leverages the ensemble network approach in their external and internal communication.
During the 2017/2018 season, the FluSight Network ensemble was one of the top-performing real-time predictive models.
“Our collaborative, team science approach highlights the ability of multiple research groups working together to uncover patterns and trends of model performance that are harder to observe in single-team studies. The field of infectious disease forecasting is in its infancy and we expect that innovation will spur improvements in forecasting in the coming years,” the team said.
“Public health officials are still learning how best to integrate forecasts into real-time decision making. Close collaboration between public health policy-makers and quantitative modelers is necessary to ensure that forecasts have maximum impact and are appropriately communicated to the public and the broader public health community.”
The research from UMass Amherst is bringing the healthcare industry closer to developing a real-time predictive model for influenza transmission.
“This paper offers a survey of which models do well, when and why, plus a sort of meta-analysis of the state of the field right now,” the team concluded.
“We have brought together some of the top flu forecasting teams in the world, and through this collaboration have enabled an apples-to-apples comparison of our different methods and results.”