Researchers at Boston Children's Hospital used OpenAI's o3 Deep Research model to identify genetic causes for 18 children with previously undiagnosed rare diseases. This represents nearly 5% new diagnoses from hundreds of genomes that had already been analyzed multiple times.
Researchers at Boston Children's Hospital and OpenAI used OpenAI's o3 Deep Research model to identify genetic causes for 18 children with previously undiagnosed rare diseases. The findings were published in NEJM AI.
Thousands of children visit Boston Children's Hospital daily, but some with rare diseases never receive a diagnosis. The Manton Center has been analyzing genomes of over 3,500 individuals globally, but conventional methods often fail to find new answers.
This study demonstrates that off-the-shelf AI models can provide practical help in diagnosing rare diseases. Although a 5% new diagnosis rate may seem small, the researchers emphasize that these cases had already been analyzed multiple times, and each diagnosis means an answer for a family.