The National Institutes of Health
From NIH Research Matters (NIH)
December 11, 2012
Scientists were able to forecast seasonal flu outbreaks using an approach common to weather prediction. The accomplishment lays the groundwork for systems to help public officials better predict and prepare for outbreaks.
In temperate regions, people become sick from influenza infections most often during winter. Dry air appears to be a factor. People also spend more time indoors together when it's colder, giving flu viruses more opportunity to spread. But beyond this general trend, our ability to make real-time predictions of the timing, duration and magnitude of local seasonal flu outbreaks remains limited.
Dr. Jeffrey Shaman at Columbia University teamed up with Dr. Alicia Karspeck of the National Center for Atmospheric Research to develop a way to more precisely predict the course of seasonal flu outbreaks. They used an approach similar to that employed by meteorologists to forecast weather. In weather prediction, new data must continually be taken into account as atmospheric conditions change. Prediction of disease outbreak, they reasoned, also needs to continually incorporate new information.
Shaman had previously developed a mathematical model of flu transmission that takes into account how humidity levels affect susceptibility to infection. In the new study, the researchers incorporated Google Flu Trends data into the model. Google researchers—working with scientists at the Centers for Disease Control and Prevention—have shown a close relationship between how many people search for flu-related topics and how many actually have flu symptoms. Google Flu Trends uses search data to estimate current flu activity in numerous countries. It also tracks activity by district, province, state or municipality.
The researchers incorporated these web-based estimates for the 2003–2008 influenza seasons in New York City into their model of influenza transmission dynamics. Their study was funded by NIH’s National Institute of General Medical Sciences (NIGMS) and National Institute of Environmental Health Sciences (NIEHS), along with the Department of Homeland Security. Results appeared online on November 26, 2012, in Proceedings of the National Academy of Sciences.
The scientists assimilated weekly flu activity estimates into the simulations. They then generated weekly forecasts using the optimized model. They showed that, by using this method, they were able to make accurate real-time predictions of flu outbreak peaks more than 7 weeks in advance of the actual peaks.
The ability to predict the timing and severity of seasonal flu outbreaks can help health officials and the general public better prepare. “Flu forecasting has the potential to significantly improve our ability to prepare for and manage the seasonal flu outbreaks that strike each year,” says Dr. Irene Eckstrand of NIGMS.
The scientists expect the accuracy of their model's predictions to rise as more years of Google Flu Trends data and more locations become available. The approach can also be adapted to develop predictions for other seasonally recurring respiratory diseases, such as respiratory syncytial virus—a major cause of respiratory infections in children—and rhinovirus, which causes the common cold.
—by Harrison Wein, Ph.D.