Skip to content

Where will Ebola spread next? Travel patterns offer clues

By the time tests confirm Ebola in a community, the response is a race against the clock. A Northeastern professor and Congolese experts built a model that’s helping health teams get ahead of the outbreak.

Health worker in a blue protective gown, face mask and gloves aims a thermometer gun at a man’s forehead from a booth at the center’s gated entrance, beside a hand-painted French sign listing visiting hours for Ebola patients.
In the Democratic Republic of the Congo, mobile phone records could help clinics prepare for Ebola before the first patient arrives. (AP Photo/Sebastien Kitsa Musayi)

How many trips have you squeezed into the past 10 days? Whether you’re in Boston or across the globe, in the Democratic Republic of the Congo, your daily routine keeps you on the move. Between seven work commutes, a couple of grocery runs, a pharmacy stop and weekend dinner with friends, you can cover a lot of ground.

Whatever your zip code, these outings bring you into contact with other people. If you’re sick but brush off early symptoms, those encounters give the infection a chance to spread. 

The stakes are higher than ever in the DRC this year, where the country’s seventeenth Ebola epidemic has become its deadliest and the second largest anywhere on record, according to Northeastern experts. The outbreak, declared May 15, 2026, has claimed 5,323 lives, with more than 10,920 confirmed cases as of November, according to World Health Organization reports. 

Containing this particular outbreak has been a perfect storm of challenges, according to Samuel Scarpino, Bouvé College of Health Sciences professor of public health and health sciences and executive director of the Institute for Experiential AI. With active conflict limiting access and a testing snafu that caused missed diagnoses, health teams found themselves playing catch-up.

In the Nov. 12 issue of the peer-reviewed academic journal Science, Scarpino and his colleagues present a simulation to help health teams predict where new Ebola cases are likely to crop up before tests officially confirm infections. 

Based on an analysis of travel patterns between communities in the DRC that were found to be the best predictors of how the outbreak spreads, the model can help officials decide where to send tests and run vaccine trials while giving clinics a chance to prepare to identify and isolate patients.

The researchers began by using mobile phone records to track trips made by people from three outbreak centers — areas where some of the earliest cases had been detected — between April 24 and May 24. Since anyone with Ebola could transport the virus, the team wanted to know which particular travel patterns could help predict where new cases would appear. 

The information, provided by an organization called FlowMinder — a leader in mobile operator data analytics — represents collective movement between areas rather than individual routes, sidestepping potential privacy concerns, Scarpino explained.

“The mobility data really tells us how many people are going between the different health zones,” Scarpino said, referring to various areas of the DRC.

Samuel Scarpino, assistant professor in Northeastern’s Network Science Institute, leans against a concrete wall outdoors.
Northeastern professor Samuel Scarpino brings the study of human movement into the fight against Ebola to help health teams protect communities most at risk. Photo by Matthew Modoono/Northeastern University

They checked which ones ended up with confirmed cases by June 1. Then, they examined how each was connected to the original centers.

Were locations geographically closer to the original hotspots at greater risk? Did actual visitor numbers matter more than physical distance? And what role did travel time from outbreak centers to each destination play?

As it turned out, all of these factors mattered, but the relationship between them was complex. 

Among communities within about 100 kilometers, or 60 miles, of the outbreak centers, those farther away were less likely to have a confirmed Ebola case by June 1.

However, geographic distance only told part of the story. Travel time mattered, too.

Say two communities are both 30 miles from an outbreak center. One has a good road people use for work commutes or shopping trips, while the other isn’t easy to reach and calls for hours of trekking over rugged terrain. Despite being at an equal distance from the initial outbreak hotspot, the place with more visitors will have more opportunities for infection to spread.

Given that travel time varies wildly in this part of the country, it offers important clues about travel patterns, Scarpino explained.

That said, knowing how long a particular trip takes doesn’t reveal how many people actually make it, he added. 

Ultimately, what mattered most was actual movement data — how many people traveled from the outbreak centers to each destination. Those patterns helped the model recreate the early outbreak more accurately than distance or travel time alone.

In fact, sometimes the pull of a certain destination trumped the distance it took to get there. Scarpino and his colleagues found a slight rise in risk 150–200 kilometers from the outbreak centers. They suggest trips to larger towns — regional hubs that attracted more visitors — could explain the uptick.

The researchers then used these patterns to build a simulation of how this virus spread and checked how closely its predictions matched actual case records. The versions that incorporated movement estimates from phone data came closest to reproducing the observed patterns.

A shared dashboard with updated outbreak data helped health specialists put the findings to work as the analysis took shape.

“The question of how knowledge of population movements has helped guide the response is an important one,” Raoul Kamadjeu, a public health specialist at the United Nations Children’s Fund, UNICEF, in Nairobi, Kenya told Northeastern Global News. His team has also used FlowMinder mobility data to inform outbreak response efforts.

“Preparedness activities have been conducted in areas identified as being at high risk,” he said. “At the same time, limited resources and the scale of the outbreaks have required us to prioritize areas already affected.”

Scarpino said the study brought together researchers, local leaders and response teams.

“It represents a new way of responding to outbreaks together as international consortiums collaborating, led by the country that is experiencing the outbreak,” he said. “And that’s a different paradigm for the field.”

Having local representatives on board can make or break the response, health experts said.

“Local leadership matters just as much as speed,” Rachel Howard, senior technical emergency health adviser at the International Rescue Committee, told NGN. “Congolese public health experts know things a dataset can’t show, for example which roads and markets move people between communities.”

The collaboration also helps the results pass muster with officials, who are more likely to act on findings they helped develop, she added. 

is a science reporter at Northeastern Global News. Email her at e.poltorak@northeastern.edu.

Katya Poltorak is a science reporter at Northeastern Global News. Email her at e.poltorak@northeastern.edu.