Influenza is a common virus that regularly targets a certain area during specific times of the year. In the United States, flu normally breaks out every winter.
Due to weather and climactic differences, however, the case may be different in another country.
Scientists at the Mailman School of Public Health at the Columbia University collaborated with the University of Hong Kong's School of Public Health of the Li Ka Shing Faculty of Medicine in a study to develop a real time operational system that can be used to predict influenza outbreaks and aid in public health decision making in subtropics.
In a study published online in the journal PLOS Computational Biology, the researchers said that in subtropical countries like Hong Kong where flu occurs differently and at varying times, timing and intensity can be predicted.
"These forecasts provide information at lead times that can be valuable for both the public and health officials," said Jeffrey Shaman, Mailman School's associate professor of Environmental Health Sciences, and the study's senior author. He added that this gives individuals the choice to get a flu vaccine, and health officials to have an idea of how many vaccines, supplies and health care personnel are needed.
The study involved data gathered from 50 outpatient clinics in Hong Kong, including lab reports between 1998 and 2013. The data was used for a test case that can generate forecasts retrospectively, on a weekly basis.
For 44 epidemics caused by flu strains for a period of 16 years, the researchers were successful in forecasting the timing and magnitude of the epidemics, at their peak levels. These included influenza A - also known as H3N2, influenza B and H1N1.
Some of the techniques used to predict local weather seen in regular nightly news were applied to the team's prediction of influenza outbreaks. Through these techniques, the researchers were able to predict the outbreak's peak timing, as early as three weeks prior to the actual occurrence. The team added that the predictions were as accurate as 93 percent.
Depending on the strength of the strains of influenza outbreak and how earlier the forecasts are made, accuracy in prediction varied. The team noted that specific strains led to more accurate predictions, compared with aggregate epidemics. Timing and peak were also more accurately predicted than onset and duration.
Using a system recognized by the CDC, Columbia University's Mailman School has been publishing weekly predictions of flu outbreaks in more than 100 temperate cities in the US since the 2013 to 2014 season. In Hong Kong, which has a non-temperate climate, 7 million people make the cities densely populated. The "global epicenter for flu" has high-quality data that the researchers were able to incorporate into their predictions.
Photo: See-Ming Lee | Flickr