Mapping Infectious Diseases
A system maps outbreaks culled from Internet news sites.
Wonder what state or country has been hit hardest by the salmonella outbreak or where the latest cases have been detected? You can find out on HealthMap, a public-health surveillance system that scours Internet news sites and other sources for real-time information on outbreaks of infectious disease, and then plots that data on a freely accessible world map.
The HealthMap software first collects information from various sources, including alerts from the World Health Organization, discussions on public-health listservs, and breaking news from thousands of websites in six different languages from around the world.
A series of text-processing algorithms then picks out the disease being reported and the location of the event, and tries to determine its relevance. For example, the software must distinguish between a breaking news item on tuberculosis and an article about a TB vaccination campaign. John Brownstein, cofounder of HealthMap and an assistant professor at the Informatics Program at Children’s Hospital Boston, says that the algorithms rate reports correctly about 95 percent of the time.
The data is plotted on a world map, with different colors indicating the most recent reports. “It’s very helpful to be able to see these things in a spatial representation–it helps us recognize when a disease is spreading, when there’s a cluster of cases, and when different cases might be related,” says Larry Madoff, a physician at Harvard Medical School, in Boston, and editor of ProMED, a public-health listserv run by the International Society for Infectious Diseases.
One of the benefits to this automated approach is the speed: local news organizations and listserv discussions often detect cases prior to official public-health channels. “We’ll pick up a case of avian influenza in Indonesia, and then a few days later see an official confirmation from the World Health Organization,” says Brownstein. The site is fully automated and is updated every hour, 24 hours a day.
World travelers can check the map to look for new outbreaks in their vacation destinations. But from a broader public-health perspective, researchers say that the system is likely to be of most use in poorer nations, which have little in the way of public-health monitoring and are often a hotbed of infectious disease. Whereas in countries like the United States, where public-health officials monitor local emergency-room visits and other indicators for signs of outbreaks, “many developing counties don’t have the capacity to do surveillance,” says Brownstein. “In those cases, the news media tends to be a good source of data.” On the downside, areas that carry the greatest burden of infectious disease, including large parts of Africa and South America, have fewer media outlets and thus the greatest gaps in coverage.
These sources become especially important when a country tries to keep public-health information under wraps for economic or political reasons, experts say. “An outbreak of cholera mandates restrictions on trade and tourism that are bad for the country, so they’ll err on the side of caution,” says Madoff. “But from a public-health perspective, it may be more important to act sooner just on the suspicion [of an outbreak].”
HealthMap’s use of so-called informal sources builds on a growing trend. “News reports, blog reports, and various other sources have become more important sources of information for public health,” says Madoff. For example, early indications of the SARS outbreak in China were evident in local discussions on ProMED, an email-based reporting system, long before the outbreak made global headlines. HealthMap uses ProMED, which was developed more than a decade ago and was the first to leverage the Internet to track infectious disease, as one of its data sources.
But along with the widening net of information used by resources like HealthMap comes concern over the accuracy of its data. “Since the information is based on reports combed from the Internet, there may be something lost in translation,” says Arnold Monto, an epidemiologist at the University of Michigan School of Public Health, in Ann Arbor. That will become an even bigger concern as HealthMap researchers begin to explore an even broader array of sources. “This is a major consideration as we move into blogs and chat rooms,” says Brownstein. “We’re spending a lot of time thinking about how valid and reliable sources are.”
HealthMap is also collaborating with ProMED, which is refereed by experts; a team of specialists monitor world news and reports from users and send them out to the listserv. Researchers aim to capture the best of both worlds, using HealthMap’s automated technology to quickly collect data and ProMED’s human experts to curate that information.
Some see HealthMap’s wide net as an advantage. “I think this is the future direction of infectious-disease epidemiology,” says Durland Fish, an epidemiologist at Yale School of Public Health, in New Haven, CT, who has previously worked with Brownstein. “This goes beyond traditional human case reporting, which we’ve done in the past,” he says. For example, the software may pick up reports from scientists working outside the public-health arena, perhaps identifying a new animal virus with the potential to jump to humans. That could conceivably give public-health officials the power to intervene before an outbreak. “We would be able to identify where this virus is likely to emerge in humans,” says Fish. “Once we get the first human case, we would know the virus has made the jump. That’s presumably what happened with SARS.”