International coalition uses the Cloud model to speed epidemic modelling and direct field resources for Ebola in eastern Congo

The response to the most dangerous outbreaks in eastern Democratic Republic of Congo begins with simple field steps: a community health worker moves between homes in the city of Butembo carrying a notebook in which they record cases of fever, deaths and recoveries, then send their daily summary via the WhatsApp app to the area supervisor. These rapid manual observations constitute the first line of defence against the spread of a rare and lethal strain, as health authorities declared an international emergency on 17 May after confirming an outbreak of Ebola virus of the Bundibugyo strain. Unlike the Zaire strain, which has an approved Ervebo vaccine, there is still no approved vaccine for the Bundibugyo strain, putting medical teams in a direct race against time to contain the infection by relying on rapid surveillance and resource allocation.

Field figures reveal a sharp operational pressure on the health system in Congo;The total confirmed cases since the start of the outbreak have reached 7,672, with deaths rising to 3,699, a case-fatality rate of 48.2 percent, while 886 patients are under isolation and 1,879 have recovered. In just twenty-four hours, teams recorded 58 new confirmed infections distributed across Ituri (38 cases), North Kivu (14 cases), Haut-Uélé (5 cases) and Tshuapa (one case), along with 23 new deaths, of which 13 occurred in the community and 10 in treatment centres, and 15 recoveries. This acceleration pushed capacity to the brink of collapse, with treatment-bed occupancy in North Kivu province reaching 94.7 percent and all beds in the cities of Butembo and Katwa exhausted.

Facing this field bottleneck, the artificial-intelligence system entered the heart of the operations room through a partnership led by the Coalition for Pandemic Preparedness Innovations in collaboration with the World Health Organization Regional Office for Africa and the National Institute of Biomedical Research in Congo, where teams began using the Cloud model from Anthropic to analyse field data and support their response to the outbreak. Dr Jean-Jack Moïmbi, director general of the National Institute, discoverer of the Ebola virus in 1976 and special adviser to the director of the African Centres for Disease Control and Prevention, summarises the battle by noting that he has fought the virus for fifty years, and although the tools have changed fundamentally, the principle remains the same: defeating Ebola requires knowing where the virus is today, not where it was last week.

The practical shift imposed by the model became evident in epidemiologists’ ability to overcome time constraints. Paul Oma, a member of the World Health Organization Regional Office for Africa team, explains that the team previously could not even consider comparing different mathematical models of disease spread, relying on a single model because of limited time, whereas today they can run multiple models simultaneously to generate forecasts that help logistics officials pinpoint exact locations for new treatment centres. In addition to spread forecasting, the Cloud Science environment enables researchers to use natural-language commands to assemble the viral genome and build a phylogenetic tree of the virus’s genetic lineages, without replacing human staff or delegating scientific decision-making to artificial intelligence, as the tools remain accelerators for analysis rather than substitutes for clinical and field judgement.

What this experiment reveals moves the artificial-intelligence dialogue in the Arab health sector from a circle of demonstrative trials to actual emergency plans.For those leading digital transformation or health-crisis management in ministries of health and hospitals across the Gulf, Egypt and the Arab Levant, the Congo case provides an operational lesson on bridging the gap between unstructured raw field data and rapid central modelling. The real challenge in managing epidemics or periods of mass pressure does not lie in the absence of data, but in the slowness of its engineering; the ability to query generative models to build multiple scenarios simultaneously reduces the time needed to decide on reallocating beds, medical staff and drug stock from days to hours, turning artificial intelligence into a critical logistical infrastructure for protecting health systems at peak times.