Since May, health teams in the Democratic Republic of the Congo (DRC) have been responding to an outbreak of Bundibugyo virus (BDBV), a rare strain of Ebola. Community health workers known as relais communautaires go door to door, offering counsel, recording recoveries and deaths, and referring people with suspicious symptoms for care. In this outbreak, nearly half of the almost 8,000 confirmed cases have died, and how quickly a person is found often determines whether they survive.
In an outbreak, a health worker's notebook is often the first and most complete record of when someone fell ill, whom they contacted, and where they went before they knew they were infected. That same contact information will later guide where a vaccine needs to go once one becomes available. But the response is limited by how quickly data moves from notebooks to labs, and then to ministries and funders deciding where resources go.
Health facilities typically send case information to districts over WhatsApp and ship biological samples to local labs for sequencing. Districts compile case data into PowerPoint slides for the outbreak coordination center, where provincial health ministry staff work late nights assembling the published situation report, or sitrep. A partnership convened by the Coalition for Epidemic Preparedness Innovations (CEPI) with the WHO Regional Office for Africa (WHO AFRO) and the Institut National de Recherche Biomédicale (INRB), among others, has begun changing that process in eastern DRC by working with Anthropic's Beneficial Deployments and Applied AI teams to apply Claude to field data, research evidence monitoring, and a scientific workbench.
At WHO AFRO's emergencies hub in Dakar, Senegal, Tendai Muza and colleagues Tamayi Mlanda and Gianni Donkor Muza built a Claude skill that pulls case and lab numbers from each health zone's PowerPoint deck, checks them against the previous day's report, flags trend changes with possible explanations, and summarizes district reports. A sitrep that once took all day can now be completed in under an hour. The team also uses Claude for analysis and disease modeling; Paul Ouma said that previously time constraints meant they could only run one model, but now they can run multiple at once and build forecasts to help logistics staff decide where treatment centers should be built.
There is currently no approved vaccine for BDBV, though Ervebo is approved for the Zaire strain. When BDBV emerged, CEPI invited proposals to advance potential vaccine candidates and related science. Polina Brangel, R&D Data Innovation Lead at CEPI, said Claude does not decide which cohorts of serological samples should be analyzed to generate evidence on Ervebo's potential cross-reactivity against BDBV — that remains a scientific judgement made by experts. What Claude does, she said, is organize complex, multi-factor data so the human eye can compare across a wide range of proposals and draw conclusions within a much shorter timeframe. CEPI also used Claude to build a dashboard tracking the actions teams need to take to advance vaccine development.
In outbreak labs, the bottleneck to results is often bioinformatics — the computer analysis of a virus's genetic data. The INRB lab has produced essentially all of the BDBV genomes sequenced from DRC cases in the current outbreak. Each sequencing machine produces millions of short fragments of genetic code, and assembling a full genome from them has traditionally required typing specialized commands into a programmer's terminal. Claude Science, an AI workbench for scientists, can be prompted in plain language to assemble the genome and build the virus's family tree, lowering the barrier to analysis that can trace new infections, estimate the outbreak's scale, and identify new variants. WHO Africa is already applying similar tools to a chikungunya outbreak, using Claude to clean and validate line lists, and next wants to use it to stitch together historical disease databases so future decisions can draw on past outbreaks.
Global health partners are beginning to see where AI tools can bring the greatest benefit during an outbreak. The Bundibugyo Ebola response still has a long way to go, though there are some promising signs of the outbreak slowing in Ituri. Response teams are helping the world learn what responsible, practical AI deployment can look like in real-world public health situations, where time is of the essence.