Flu forecasts may soon become a news staple.
Forecasting the flu. I’m Bob Hirshon and this is Science Update.
Soon, you may be able to check a flu forecast along with the weather report. This according to environmental health scientist Jeffrey Shaman, of Columbia University’s Mailman School of Public Health. His team used data from Google Flu Trends, a tracker of flu-related Web searches that’s a surprisingly good indicator of actual cases. By combining weather forecasting techniques and models of flu transmission, Shaman and his team could predict the course of an outbreak.
So the specific details of the where and when, when it’s going to peak, how many cases there may be, what’s the duration of the outbreak going to be, and what’s the total number of cases going to be.
They tested it on flu patterns from 2003 through 2008. Shaman says the forecasts could be used not only to warn the public, but also to alert medical authorities for a potential flood of patients. I’m Bob Hirshon for AAAS, the Science Society.
Making Sense of the Research
Partly cloudy with a 50 percent chance of flu? That's not quite how it would sound. But there soon could be flu risk forecasts added to the predicted temperature, humidity, wind chill, UV index, pollen count, and air quality measures that you can now track daily.
Tracking cases of the flu directly is difficult, since official reports can take up to a week and go through many different national, state, and local agencies. But the researchers tapped into a surprisingly useful indirect measure of flu cases: Google Flu Trends. Google Flu Trends tracks how many people search for terms like “flu medicines,” or “flu symptoms.” A report in the prestigious scientific journal Nature showed that these Internet searches correlate very closely with actual flu cases since 2003.
Google Flu Trends covers the past and the present (up to the minute!), but Shaman's team wanted to predict the future. So they combined this data with computer models of how flu tends to spread, and used weather forecasting techniques to make predictions about the intensity of outbreaks in different regions. Their model is now able to successfully “predict” past outbreaks based on the information that was available beforehand, which means it should be similarly successful at predicting future outbreaks.
Although we all know that when the weather report calls for rain, we take an umbrella, it's less clear exactly how people would behave differently if they knew a flu outbreak was likely. But it's possible that the awareness could lead to better preventive habits, like hand-washing, avoiding crowds, and so on. The projections also would be useful to hospitals, which can be flooded with emergency room visits during a major flu outbreak.
One way this is different from weather forecasting is that the forecast itself could affect the outcome. In other words, if it's supposed to rain, taking your umbrella doesn't make rain less likely—it just makes you prepared for it. But if enough people pay close attention to flu forecasts and take precautions, it's possible that predicted outbreaks won't be as severe as expected, which would make the forecasts less “accurate.” How scientists would deal with that problem remains to be seen, but it's not such a bad problem to have.
Now try and answer these questions:
- What is flu forecasting?
- What information does the flu forecast use? What kind of modeling techniques go into it?
- Why is it scientifically valid to use Internet searches as one way to track and measure flu outbreaks?
- What are some of the possible benefits of a flu forecast?
The lesson, Air Masses helps develop an understanding of air masses and the role they play in weather and climate using models.
Another Science NetLinks lesson, Sanitation and Human Health, takes a look at the impact of improved sanitation on human health while Epidemiology: What's My Hypothesis? demonstrates how descriptive epidemiological clues can be used to make educated guesses as to what might be the cause of a disease.