OpenAI and Boston Children’s researchers use AI to reanalyze rare-disease cases

OpenAI published a study with researchers from Boston Children’s Hospital, Harvard University, and the Manton Center for Orphan Disease Research on AI-assisted rare-disease analysis.

The study used OpenAI o3 Deep Research to analyze de-identified clinical and genomic information from 376 previously unsolved cases. The cases had already gone through earlier expert review and testing.

Researchers asked the model to propose possible molecular explanations and provide supporting evidence. The model did not make diagnoses or clinical decisions. Its outputs were reviewed by experts using established clinical genetics standards.

After expert review, additional testing, and clinical confirmation, physicians established diagnoses in 18 cases. That equals an additional diagnostic yield of 4.8% across the reviewed cases.

The study was published in NEJM AI on June 18, 2026. OpenAI said the results show that AI-assisted workflows may help experts revisit unresolved rare-disease cases as new genetic evidence and medical literature become available.

Source: OpenAI — https://openai.com/index/diagnose-rare-childhood-diseases/