Could Twitter prevent epidemics? Using the Internet to survey public health

Most of us have turned to the Internet when we’ve felt under the weather at one point or another. We Google our symptoms to self-diagnose and sometimes we even tweet our distress, perhaps hoping to gain a little sympathy. Researchers have been looking into how this information we put out can be interpreted to map both disease and the spread of information about disease.

In January, two studies were published that assessed the effectiveness of such methods in quickly and accurately tracking illness. One reported on tracking the cholera outbreak in Haiti, which appears to have begun with the death of a man in October of 2008. Through HealthMap, which collects information on illnesses around the world from online sources including news media and blogs, researchers collected 4,697 reports, in addition to 188,819 Tweets within the first 100 days of the cholera outbreak. The researchers found that they could make predictions about where the disease was spreading and in which communities the next outbreak was likely to occur weeks sooner than with traditional survey techniques, like reports from hospitals and health clinics. In future outbreaks of disease, these techniques could be applied to provide more timely and effective treatment and vaccination efforts.

The other study focused on possible practical applications of Google Flu Trends by medical professionals. Google Flu Trends, which was launched in late 2008, provides a world map with minimal, low, moderate, high and intense flu activity for more than 20 different countries. It also features graphs of current flu trends compared to past years. Johns Hopkins researchers found that over 21 months, when data gathered by Google showed an increase in searches for influenza-like symptoms, it correlated with an increase in people coming into a busy emergency room with those same symptoms, meaning that Google Flu Trends could be used to warn health care providers and allow them to prepare for an influx of patients.

The topic has also come up among those in the business of food safety. Speaking at this year’s Global Food Safety Conference in February, Frank Yiannas, vice president of food safety at Walmart, said that reports of outbreaks are often released too late. He believes that social media will change how food safety is managed.

The ideas behind this method of information gathering can be traced back to 2004, when Dr. Gunther Eysenbach coined the terms ‘Infodemiology’ and ‘Infoveillance’ as he started exploring the possibilities of using the Internet to gather information about public health. Infodemiology is the tracking of the spread of symptoms as well as the spread of knowledge about disease. Infoveillance is the type of surveillance methodology that gathers this information.

More recently, Eysenbach looked at what Twitter revealed about the H1N1 virus -- or swine flu -- outbreak in 2009. He not only tracked what people were saying about the virus, but also how it correlated with specific events, such as when vaccinations began in the US. A presentation of his results can be found here.

As these methods are being employed in more areas, the question of their accuracy persists. As more studies show that social media and Internet searches can be useful, the information should still be taken with a grain of salt. For example, Flu Near You provides maps of influenza activity in the United States based on Centers for Disease Control and Prevention (CDC) reports, Google Flu Trends, and data submitted directly by users. A quick comparison of the data sets shows little consistency. While it is not likely one report will say flu activity is high in a particular state while another says it is low, the measures of activity by state rarely match up exactly. Overall, it seems these methods are useful for monitoring trends, and some may be more accurate than others in pinpointing exact locations of outbreaks.

When we go on the Internet seeking information about our personal health or to tweet our symptoms or share information we might have heard about a particular illness, not thinking that anyone other than our friends would see or care, we could be contributing to the greater public health. As these information-gathering techniques continue to evolve, search engines and social media could play a role in preventing outbreaks before they begin.