Researchers Are Closing In On a Method to Predict Flu Outbreaks
In the near future, you may be able to hear a routine "flu forecast" on the evening news just as weather forecasts are issued each day. A team of scientists working on a flu forecast model recently announced they're a step closer to producing a model that can reliably warn consumers of flu trends even seven weeks in advance.
Having a method to predict flu outbreaks could help curb the spread of the virus during peak times of the year and could potentially save many lives during a severe pandemic.
"I can envision this being part of the local weather forecast in the same way you have pollen counts and pollution levels," lead author Jeffrey Shaman, an assistant professor of environmental health sciences at Columbia’s Mailman School of Public Health, told Take Part.
The paper, which appears in the Proceedings of the National Academy of Sciences, is the latest in a series of work by Shaman and colleagues at the National Center for Atmospheric Research. The research is unusual because it applies weather forecasting models to an infectious disease.
Flu is particularly well-suited to this kind of forecasting. There are few other illnesses as seasonal as flu. In the United States, it's basically a wintertime ailment. However, the authors of the paper note, the flu peak each year can vary dramatically. In some years, flu cases peak before the holidays and in other years, flu peaks as late as April.
In a 2009 paper, Shaman showed how important environmental factors are on flu transmission and persistence. Flu epidemics tend to follow very dry weather. But in the new study, Shaman and co-author Alicia Karspeck, of NCAR, borrowed a mathematical model used in weather forecasting and combined it with real-time information gathered from Google Flu Trends. Google Flu Trends shows the number of flu-related search queries in a given region and estimates outbreaks based on that demand for flu information.
In the study, the researchers used the model to look, retrospectively, at flu during the winters of 2003-04 and 2008-09 in New York City. The model accurately predicted the peak timing of the outbreak more than seven weeks in advance. While a weather forecast predicting conditions seven weeks out is not likely to be very accurate, the flu forecast model produced a surprising level of accuracy.
"One of the things about a weather forecast is it degrades the farther out you go," Shaman says. "Weather forecasts degrade at a certain rate because of the complexity of weather. Flu trends are simpler."
The researchers' next step is to create models for various regions of the country because flu cases peak at different times around the country. But Shaman says he thinks a usable flu forecasting system could be available to public health officials in about a year or two.
Eventually, he says, flu forecasts could be made available to the public.
"We can generate forecasts very similar to what you get in weather," he says "There is a 40 to 60 percent chance that flu will peak in six weeks -- that would be a strong forecast. There is a 10 percent chance that flu will peak in five weeks would be a weak forecast."
Consumers love to hate weather forecasts, and it might take some time for people to buy into a flu forecast. But Shaman says he thinks consumers will pay attention. Flu infects millions of Americans each year and, on average, causes about 35,000 deaths each year. The illness accounts for many lost school days and time away from work. A flu forecast could prompt people to be more diligent about prevention measures, such as frequent hand-washing, staying home from work when they feel symptoms coming on and even wearing masks to avoid exposure, the authors note.
"If we show the forecasts are reliable and people get a sense that they are, I think they'll respond to the information sensibly," he says. "The forecasts even have enough lead time that it can be an effective reminder for people who haven't been vaccinated to go get the vaccine."
Public health officials and medical professionals would benefit from a flu forecast by stocking up on anti-viral medication and scheduling extra personnel in hospitals and clinics to care for more patients, he says. In a severe pandemic, city officials could use the information to close schools to thwart transmission.
Question: Do you think it would be helpful to have weekly "flu forecasts" during the winter months? Tell us what you think in the comments.