Could mobile phone data help predict where Ebola will spread next?
An analysis of the early stages of the Bundibugyo virus outbreak in the Democratic Republic of the Congo shows how population movement data could help predict disease spread.
- 18 November 2026
- 6 min read
- by Linda Geddes
At a glance
- Researchers have combined infection records with anonymised mobile phone data and other information to reconstruct the early spread of Bundibugyo virus in the Democratic Republic of the Congo.
- They found that population movements along major transport routes may have played a key role in the outbreak’s early geographical spread, and that mobility data could strongly predict which areas subsequently reported infections.
- The findings suggest that such data could help anticipate where infections might emerge during future outbreaks, enabling health authorities to strengthen surveillance, deploy diagnostic equipment more efficiently, prepare healthcare facilities and engage with communities sooner.
Human movement along major transport routes may have played a key role in the early spread of Bundibugyo virus in the Democratic Republic of the Congo (DRC), new research suggests.
By combining records of confirmed infections with anonymised mobile phone data and other information, researchers have reconstructed the first few weeks of the ongoing Ebola outbreak there.
They found that data on people’s movements could strongly predict which areas subsequently reported infections, potentially providing an additional tool for health authorities responding to such outbreaks.
The researchers also examined how quickly the outbreak grew during its early stages, finding that the number of cases approximately doubled every ten days between early May and 1 June 2026.
While this is towards the faster end of growth rates recorded during previous Ebola outbreaks, factors such as delayed recognition of the outbreak, difficulties tracing contacts, high population mobility and insecurity could have contributed to this rapid growth, rather than necessarily indicating that the virus itself is more transmissible.
What happened during the first few weeks of the Bundibugyo virus outbreak?
When the Ebola outbreak in DRC was first declared on 15 May 2026, there were already signs that Bundibugyo virus had been circulating for several weeks, with infections reported across multiple health zones. However, it was uncertain how quickly the virus was spreading, and which areas were most at risk.
Dr Dav Ebengo at the Institut National de Recherche Biomédicale (INRB) in Kinshasa, and colleagues set out to reconstruct the outbreak’s early development, identify factors driving its geographical spread and explore whether population movement data could help predict where infections might emerge next.
Their aim was to generate information that could help health authorities assess risks and guide their ongoing response.
To investigate, they combined detailed records of confirmed infections with anonymised mobile phone data on population movements, and information on travel times, healthcare access and demographics.
While the principle of using such mobility data to predict the geographical spread of Ebola or other infectious diseases is not entirely new, “its application to this virus, at this stage and at this speed, is,” said study co-author Dr Pierre Akilimali at the Institut National de Santé Publique in Kinshasa, DRC.
They found that population movement data collected in March 2026, before the outbreak was recognised, were highly predictive of which health zones subsequently reported infections. Had the analysis been run on the day the outbreak was declared, this could have providing health authorities with valuable time to prepare, the researchers said.
Another important finding was that population size alone did not predict which health zones would report infections. “What carries the signal is the connection between places, not their size,“ Akilimali said.
The study, which was published in Science, also found that publicly available travel-time data could identify areas at risk with considerable accuracy, suggesting that useful assessments might be possible even where mobile phone data are unavailable.
How could data about people’s movements be used to improve outbreak responses?
The findings suggest that mobility data could help anticipate where infections might emerge during future outbreaks, enabling health authorities to strengthen surveillance, deploy diagnostic equipment more efficiently, prepare healthcare facilities and engage with communities sooner.
“The practical advantage is lead time, which is the currency of outbreak response,” said Akilimali.
Indeed, areas identified by their models as being at considerable risk in early June subsequently reported infections in July, providing around five to six weeks’ advance warning – long enough to prepare laboratory services, train health workers to investigate suspected cases, strengthen infection prevention measures, and increase surveillance along transport routes.
“It changes surveillance from a pursuit into a placement exercise: you are no longer waiting to be told where the outbreak has arrived,” Akilimaki said.
Have you read?
Could this approach be used more widely?
Looking ahead, Akilimali believes a similar approach could help anticipate the geographical spread of other infectious diseases, including measles, cholera, mpox, meningitis and Marburg virus disease. “The method is generic; the task now is to make it routine rather than exceptional,” he said.
He added that it could be particularly valuable for vaccine-preventable diseases, where identifying communities at greatest risk could help authorities decide where to prioritise vaccination campaigns and how best to allocate available doses.
The technical requirements are relatively modest. The researchers used aggregated, anonymised mobile phone data supplied by Vodacom Congo and processed by the Flowminder Foundation, alongside publicly available information about population distribution and road networks. The analysis could be performed on ordinary computers.
However, Akilimali stressed that agreements with mobile phone operators need to be established before an outbreak begins, as negotiating data access during an emergency could cost valuable weeks.
For the current Bundibugyo virus outbreak, “one point deserves to be made plainly,” Akilimali said. “With no licensed vaccine and no licensed therapeutic against Bundibugyo virus, anticipation is the one countermeasure that can be applied before the virus arrives.”
He added: “If candidate vaccines are deployed, these same maps tell you where to put them. That, in our view, is the argument for investing in this kind of work between emergencies rather than during them.”
What about privacy concerns?
Importantly, the researchers did not receive information that could be used to identify individuals, such as names, phone numbers or individual movement histories. Instead, they used aggregated estimates of how many mobile phone users moved between different areas over periods of weeks.
“People are right to be wary of being tracked, and that is precisely why privacy has to be built into the design of these arrangements rather than negotiated in the middle of an emergency,” Akilimali said.
Beyond the potential of mobility data to predict where infectious diseases might spread, the study also highlights the growing capacity of Congolese scientific institutions to investigate outbreaks as they unfold.
“Twenty years ago, samples left this country and answers came back months later,” said study co-author Professor Placide Mbala-Kingebeni of the Institut National de Recherche Biomédicale and Africa CDC. “This time, the sequencing, the diagnostics and the epidemiological analysis were led from Kinshasa and Bunia while the outbreak was still growing. This local capacity will support future epidemic prevention and response.”
More from Linda Geddes
Recommended for you