International health teams are using Anthropic’s Claude tool to accelerate the response to an outbreak of a rare strain of the Ebola virus in eastern Democratic Republic of the Congo, where no vaccine has yet been confirmed as approved for this strain. The use focuses on turning scattered field data into reports and analyses that can help response teams identify where infections are spreading and direct resources.
The initiative involves the Coalition for Epidemic Preparedness Innovations (CEPI), the World Health Organization Regional Office for Africa (WHO AFRO), the Democratic Republic of the Congo’s National Institute for Biomedical Research (INRB), along with Anthropic teams focused on public-benefit deployment and applied artificial intelligence.
From Field Notebooks to a Situation Report
Community health workers record cases of fever, recovery, and death every day, and send their notes via WhatsApp to district officials. Case data and laboratory results are then compiled into PowerPoint presentations that move between districts, response coordination centers, and the health ministry before being converted into the situation report known as a sitrep.
A team at WHO AFRO developed a custom skill for Claude that extracts case and laboratory figures from district presentations, compares them with the previous day’s report, identifies changes in trends, and summarizes district reports. According to the article, the time required to prepare the report fell from a full working day to less than an hour.
The tool is also used to analyze data and model the spread of the disease. Paul Ouma of the WHO AFRO team said that teams are now able to run multiple models at the same time, instead of relying on a single model because of time constraints, helping logistics teams assess where to establish treatment centers.
Supporting Vaccine Research and Genetic Analysis
There is currently no approved vaccine for the Bundibugyo strain, while the Ervebo vaccine is available for the Zaire strain. CEPI used Claude to organize complex data and compare proposals related to vaccine development, in addition to building a tracking dashboard for actions required from research and development teams.
However, the tool does not select serological study samples or determine scientific conclusions; Polina Brangl of CEPI explained that these judgments remain the responsibility of experts, while Claude’s role is limited to organizing data to accelerate human review.
At the INRB laboratory, Claude Science, a work environment for scientists, is used to assist in assembling the viral genome and building a genetic relatedness tree from millions of fragments produced by sequencing devices. This analysis can support the tracing of chains of infection, estimation of the outbreak’s scope, and detection of new variants.
What Is Changing in Practice?
The main change is not replacing experts, but reducing the time between collecting data and turning it into usable information. This is important in outbreaks where detecting a case before its condition deteriorates may affect the chances of survival, just as accurate contact data may help direct vaccines or protective measures when they become available.
WHO Africa has also begun using Claude to clean and verify lists of chikungunya outbreak cases, and plans to link historical disease databases so that decisions in future crises can draw on lessons from previous outbreaks. Nevertheless, the quality of the outputs remains linked to the accuracy of field data, and medical and scientific decisions still require specialized human review. The article indicates that containment of the Bundibugyo outbreak has not yet been completed, despite indications that it is slowing in Ituri.