Researchers are working to develop new ways of using Twitter messages, Facebook interactions and Google queries to predict surges in influenza cases, gain insights into how viruses spread and monitor other public health trends. But a key challenge to using social media to track flu cases is that online chatter about flu activity gets mixed in with actual messages about people who have been infected with the virus, which can skew results.
Now researchers at Johns Hopkins University have devised a new method for screening tweets that delivers real-time data on flu cases and distinguishes between general conversations about the illness and actual flu infections. As Reuters reported:
To solve the problem, [researchers] developed a screening method based on human language-processing technologies that only delivers real-time information on actual flu cases and filters out the rest of the chatter on the public tweets in the United States.
The researchers at the Baltimore university tested the system by comparing their results with data from the U.S. Centers for Disease Control and Prevention.
In late December,” [Mark Dredze, PhD, an assistant research professor in Johns Hopkins’ department of computer science] said on Thursday, “the news media picked up on the flu epidemic, causing a somewhat spurious rise in the rate produced by our Twitter system. But our new algorithm handles this effect much better than other systems, ignoring the spurious spike in tweets.”
Previously: Tracking sales of over-the-counter medicines to predict disease outbreaks, Study shows Google Flu Trends data, patient spikes at emergency departments closely correlated and Mining Twitter data to track public health trends
Photo by anna gutermuth