Big Data Takes on Dengue Fever

A School of Public Health study uses cell-phone records to predict the spread of dengue fever.

The use of mobile phone data (left) proved more accurate for estimating the spatial spread of dengue fever than a traditional diffusion model (right).
Graphic from Impact of human mobility on the emergence of dengue epidemics in Pakistan, published by PNAS

Cell phones automatically keep track of the nearest cell tower, which is why phone records can be used to determine where people were at a particular time—a principle already employed to generate alibis and convictions in the legal world. Now, researchers at the Harvard T.H. Chan School of Public Health (HSPH) have found a potentially life-saving use for these cell-phone records: to predict the spread of epidemic diseases.

By combining the phone records of around 40 million Pakistani mobile-phone subscribers (representing 22 percent of the country’s population) with climate data, the Harvard group was able to predict the spreading patterns of dengue fever, a potentially deadly tropical disease endemic in that nation. The investigators matched the trajectory of the dengue virus in a 2013 epidemic to the traveling pattern of mobile users, and found that the latter could predict the former fairly accurately. “This provides a more accurate forecast of where dengue is going to [appear] as compared to other approaches that use environmental variables or proxies of human movements like road networks,” says Caroline Buckee, senior author of the study and an assistant professor of epidemiology. For low-income countries without well-developed road networks, mobile data can be particularly helpful for tracking the spread of epidemics, adds Amy Wesolowski, a postdoctoral research fellow and lead author of the study.

Forecasting outbreaks is crucial for controlling the spread of dengue fever, which has no vaccine or treatment. Preparedness—tracking when and where it will happen—is “one of the only things you can do,” says Buckee. Because the disease spreads mainly in urban areas, where cell-phone penetration tends to be higher, dengue lends itself particularly well to tracking by proxy using mobile networks. According to Wesolowski, the authors are working to present the findings to Pakistan’s health officials. 

The Harvard study represents the largest data set of mobile-phone records ever analyzed to estimate human mobility. It was enabled by a collaboration with Telenor, the second-largest provider of mobile telecommunication services in Pakistan, and the researchers went to great lengths to protect the privacy of the company’s subscribers. The cell-phone records were anonymized, Buckee says, in the hope that the study will set a good precedent for future collaborations between health researchers and the telecommunications industry.

“Impact of human mobility on the emergence of dengue epidemics in Pakistan” appeared online on Monday in PNAS. For more about the use of big data in public health and other realms, read Why “Big Data” Is a Big Deal, a feature article that appeared in this magazine in 2014.

Read more articles by: Zara Zhang

You might also like

“It’s Tournament Time”

Harvard women’s basketball prepares for Ivy Madness.

A Harvard Agenda Shaped by Speech

The work underway in the Faculty of Arts and Sciences

Dialogue, not Debate

American University’s Lara Schwartz, J.D. ’98, teaches productive disagreement.

Most popular

alt text here

AWOL from Academics

Behind students' increasing pull toward extracurriculars

Lola Mullaney, Coach Carrie Moore, and Elena Rodriguez

“It’s Tournament Time”

Harvard women’s basketball prepares for Ivy Madness.

View of Harvard University campus from the Charles River

Why Americans Love to Hate Harvard

The president emeritus on elite universities’ academic accomplishments—and a rising tide of antagonism

More to explore

Winthrop Bell

Brief life of a philosopher and spy: 1884-1965

Talking about Talking

Fostering healthy disagreement

A Dogged Observer

Novelist and psychiatrist Daniel Mason takes the long view.