Compared to trending Twitter topics, epidemiology is a slow-moving discipline: even during a sudden outbreak of disease, it can take weeks to identify and confirm each case. Now, a growing band of researchers is advocating for greater adoption of tools such as Google News, Twitter or Facebook to help track epidemics and deploy medical aid more quickly. “Social media is here to stay and we have to take advantage of it,” says Taha Kass-Hout, Deputy Director for Information Science at the Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia.

Kass-Hout will be arguing his case for greater adoption of these tools on 16 February at the International Conference on Digital Disease Detection at Harvard Medical School in Boston, Massachusetts. The meeting aims to bring together leading figures in technology and epidemiology to discuss how ‘informal data’ such as tweets and texts can radically change disease surveillance.

Not everyone in the public-health community is ready and willing to use the data. Andrea Dugas, a research fellow in emergency medicine at the Johns Hopkins Hospital in Baltimore, Maryland, says that practitioners regard informal-data tools with “a degree of healthy scepticism”, adding that more research is needed to prove their reliability before they become a core part of public-heath planning.

Is real-time epidemiology just a tweet away? Credit: M4OS Photos / Alamy

Another challenge to the field is that recent studies have demonstrated only correlative relationships between informal data and disease incidence. For example, Dugas and her colleagues have shown that internet searches for words relating to influenza, monitored with a web platform called Google Flu Trends, tallies with subsequent hospital visits1. “You see increased search activity and a very short time later, people come to the emergency department,” says study co-author Richard Rothman, an emergency-medicine physician at the Johns Hopkins School of Medicine.

But public-health officials would find such data more useful if they could be used to predict outbreaks, and there is little evidence that these tools can do that consistently and reliably, says Rothman. To address this issue, researchers (including Rothman and Dugas) are planning larger studies to explore how tools such as Twitter and Google Trends, might hold up across different disease types and broader geographical regions to forecast emerging outbreaks.


Rumi Chunara, an engineer studying biological sensors at Harvard Medical School, says that informal data should be used as a complement to conventional surveillance, particularly in populations that may have few health resources but plenty of access to mobile phones. Chunara used Twitter-generated data to track the 2010 cholera outbreak in Haiti2, using data provided by HealthMap, a team based at Children's Hospital Boston that aggregates informal data on diseases around the world to produce an interactive map of emerging outbreaks, and the major sponsor of this week’s conference.

Designed to continuously scan blogs, tweets, official surveillance data. news sites and RSS feeds, as well as processing user-submit ted reports, HealthMap disseminates information in ten languages, and is collaborating for one of its project s with the US Department of Health and Human Services to map seasonal influenza and H1N1 in the United States.

It can be difficult for automated software to extract a useful signal from the noise, however. So Nigel Collier, a computational linguist at the National Institute of Informatics in Tokyo, Japan, is developing ways to filter the deluge. Collier’s brainchild, BioCaster, is an information portal similar to HealthMap that aggregates conventional and unconventional data across multiple languages to detect early warning signals. Like HealthMap, the information is distributed free of charge, but has a greater focus on Asian news sources and search engines, such as Baidu and Soso.

Collier says that the CDC, the World Health Organization, the European Centre for Disease Prevention and Control and the Ministry of Health in Japan have used BioCaster for a variety of purposes, including tracking of pandemic H1N1 and of cholera in Haiti.

Yet even as real-time epidemiology moves closer to reality, faster health-care delivery will always depend on financial and medical resources, cautions David Fisman, an epidemiologist at the Dalla Lana School of Public Health at the University of Toronto in Canada. “Awareness of outbreaks doesn’t necessarily change the pace or extent of intervention.”

John Brownstein, an epidemiologist at Harvard Medical School and Children’s Hospital Boston, who organized the conference and is a co-founder of HealthMap, agrees that one of the field's key goals is to demonstrate that surveillance can be translated into effective healthcare. “It’s great to know what’s taking place on the ground,” he says. “But where ’s the impact?”