Using AI to help physicians diagnose rare genetic diseases affecting children

Researchers from a children's hospital and university used an AI reasoning model to reanalyze 376 previously unsolved cases of rare genetic diseases in children. The model generated evidence-based hypotheses about potential diagnoses, which experts then reviewed and confirmed through established clinical processes. This approach identified diagnoses in 18 additional cases, an yield of 4.8%, even though these cases had already been examined by multiple specialist teams and pipelines. The work demonstrates how AI can help surface new diagnostic leads as scientific knowledge about genes and disease evolves, though the model itself never made clinical decisions or diagnoses.