FDA launches agencywide AI tool
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Tiffany Herring/Axios
The Food and Drug Administration on Monday is launching an agencywide generative AI tool aimed at helping scientific reviewers, investigators and others streamline their work, according to details shared with Axios.
Why it matters: The model, dubbed "Elsa," is being rolled out weeks ahead of schedule and is designed to help with tasks such as summarizing adverse events to support drug safety profiles, and to generate code for databases.
The big picture: The FDA said the technology was built in Amazon Web Services' GovCloud environment, which is designed to host sensitive government data. It isn't being trained with proprietary data submitted by drug and device manufacturers, according to a summary shared ahead of the public rollout.
- The FDA did not detail specifically what information the model is being trained on, but said it offers a "secure platform for FDA employees to access internal documents."
The FDA said it's using the technology to review clinical protocols and speed up scientific evaluations. It soon will be used to identify high-priority targets for inspections, though it did not specify what kind.
Zoom out: Public health experts told Axios last month that using more AI within the FDA is a good idea, but questioned how secure the data will be and how quickly the technology can really be integrated into the agency's workflow.
- Stat on Monday reported some FDA employees, speaking anonymously, were concerned about the rollout, in part because they said it's so far only been piloted to summarize texts.
- Health and Human Services didn't respond to a request for comment on the reported concerns.
