CDC awards Carnegie Mellon University $3 million for flu forecasting
Researchers at Carnegie Mellon University said this flu season has the potential to be severe.
They should know.
The Centers for Disease Control and Prevention this week named CMU as an Influenza Forecasting Center of Excellence, a five-year designation that includes $3 million in research funding.
Its Delphi Research Group, which is devoted to epidemic forecasting, has proven to be the most accurate of all the research groups participating in CDC’s FluSight Network, an initiative that encourages outside academic and private industry researchers to forecast the timing, peak, and intensity of the flu season.
“We’re pleased as punch, not just because we got awarded this money, but because epidemic forecasting has become a real thing, and it wasn’t a real thing five years ago,” said Roni Rosenfeld, head of CMU’s Machine Learning Department and leader of its epidemic forecasting efforts. “When the CDC started their first competition, it was 2013, I think there were seven submissions. Today there are about 40 every year.”
CMU uses two methods to generate flu forecasts. One uses machine learning and computational statistics to make predictions based on both past patterns and input from the CDC’s domestic flu surveillance system. The other bases its predictions on the judgments of human volunteers who submit weekly predictions.
CMU’s methods provide fairly accurate flu forecasting data.
While not at 100% because of outside factors such as doctors reports being delivered late to the CDC, they fall within a 10% error rate, which researchers consider “quite good.” For example, if the actual number is 5%, CMU said its methods will predict the flu as close as 4.5 or 5.5%.
“You collect the forecasts that we’ve made, and you look at how off they were, and you average overall a bunch of different locations in which we made forecasts and times and you get some kind of sense of how accurate we are,” said Ryan Tibshirani, co-leader of the Delphi Research Group.
The additional funding will expand CMU’s existing forecasting research, enable the university to initiate studies on how to best communicate forecast information to the public and to leaders, and support efforts to determine how forecasting techniques might apply to pandemics.
The CDC estimated that from October 2018 through May, there were 37.4 million to 42.9 million reported flu illnesses in the United States.
As for this season: “We, as a research group, still don’t have much we can say about the current flu season, but it has been pointed out by others that the flu season in the Southern Hemisphere, the flu season of the last six months, has been particularly severe,” Rosenfeld said. “So, there’s some reason to think it may be severe this year.”
To see the research visit: https://delphi.midas.cs.cmu.