Context
Telefonica R&D, in collaboration with the UN, Harvard University, and the Data Pop Alliance.
The project was initially developed in response to the Zika outbreak, in collaboration with Harvard, the WHO, and the United Nations. It was subsequently adapted for the Ebola outbreak, and later revisited as a potential tool during Covid-19. Although the system was technically feasible across all three, a decision was made not to pursue it further due to privacy concerns around the use of mobile network data at scale.
Not all countries reached the same conclusion. Taiwan, for example, went ahead and implemented a comparable system during the Zika outbreak.
Problem
During the Ebola outbreak, it took six weeks to identify the source of the epidemic. By that point, the virus had already reached Conakry, a city of over two million people. Existing epidemiological surveillance relied on clinical reports, which only emerged after symptoms appeared and patients sought care — systematically too late to contain early spread. The business problem for Telefonica was demonstrating that its network data assets had real public health value, building the case for health partnerships with governments and international organisations.
My role
I led the product and service design: translating complex mobility modelling outputs into tools that public health officials could act on in real-time outbreak situations and worked together with the Telefonica R&D research team who worked together with epidemiologists at Harvard and public health partners at the UN.
Approach
Mobile networks generate a continuous, anonymised record of where people are and how they move. Call detail records at scale reveal mobility patterns that are invisible to clinical surveillance. I worked with epidemiologists to design two applications: retrospective source detection (tracing movement patterns back to identify outbreak origin, cutting detection time from weeks to hours) and forward spread prediction (modelling likely case cluster locations before they appeared in clinical data). A third application used the same infrastructure to deliver targeted SMS health guidance to people in identified high-risk areas — without requiring smartphones or internet access.
Outcome
- Possible outbreak source detection time reduced from weeks to hours
- Collaboration with WHO, UNICEF, Harvard, and the Data Pop Alliance positioned Telefonica as a contributor to global health infrastructure
- Most effective in early-stage outbreaks where infection is still contained to a small proportion of the population. At high prevalence, as seen with Covid-19, broad public health measures become more effective than mobility-based targeting.
