Public healthPhone call patterns reveal emerging disease outbreaks

Published 18 October 2010

People who come down with a fever or full-blown flu tend to move around less and make fewer calls late at night and early in the morning; a trial shows that monitoring the calling patterns of individuals in a trial group correctly identified flu victims in the group 90 percent of the time; epidemiologists say this method may be applied nation-wide to convey telltale signatures of illness to doctors and agencies monitoring new outbreaks

Your cell phone could be a key tool in the fight against disease by relaying a telltale signature of illness to doctors and agencies monitoring new outbreaks.

This technology is an early warning system,” says Anmol Madan of the Massachusetts Institute of Technology, whose team concluded that you can spot cases of flu by looking for changes in the movement and communication patterns of infected people.

Jim Giles writes that epidemiologists know that disease outbreaks change mobility patterns, but until now have been unable to track these patterns in any detail. So Madan and colleagues gave cell phones to seventy students in an undergraduate dormitory. The phones came with software that supplied the team with anonymous data on the students’ movements, phone calls and text messages. The students also completed daily surveys on their mental and physical health.

A characteristic signature of illness emerged from the data, which was gathered over a 10-week period in early 2009. Students who came down with a fever or full-blown flu tended to move around less and make fewer calls late at night and early in the morning. When Madan trained software to hunt for this signature in the cell phone data, a daily check correctly identified flu victims 90 percent of the time.

The technique could be used to monitor the health status of individuals who live alone. Madan is developing a smartphone app that will alert a named contact, perhaps a relative or doctor, when a person’s communication and movement patterns suggest that they are ill.

Giles notes that public health officials could also use the technique to spot emerging outbreaks of illness ahead of conventional detection systems, which today rely on reports from doctors and virus-testing labs. Similar experiments in larger groups and in different communities will have to be done first though.

Leon Danon at the University of Warwick, United Kingdom, is negotiating with the ministry of health of a northern European nation about a project that would combine the anonymous cell phone records of around 10,000 people with their health records to produce signatures of disease from a larger population.

Researchers will need to think hard about the causes of the changes they see in the cell phone data, says Leon Danon at MIT, who is working with Danon. Eagle looked at cellular data from a series of cholera outbreaks in Rwanda between 2006 and 2009. He saw a clear reduction in people’s movement, which may have been due to the disease. But the outbreak was caused by floods, which also limited mobility. Distinguishing between the two possible causes on the basis of phone data alone was impossible, he says.

Madan presented his paper last month at the International Conference on Ubiquitous Computing in Copenhagen, Denmark.

—Read more in Anmol Madan et al., “Social sensing for epidemiological behavior change,” paper presented at the Proceedings of the 12th ACM international conference on Ubiquitous computing, Copenhagen, Denmark (26-29 September 2010) (http://doi.acm.org/10.1145/1864349.1864394) (sub. req.)