AI helped diagnose 18 children whose rare diseases had stumped doctors (www.nbcnews.com)

Boston Children’s Hospital has reported significant progress in diagnosing rare diseases in children, leveraging AI tools developed by OpenAI. In a groundbreaking study published in NEJM AI, the OpenAI o3 Deep Research model was used to analyze the genomes of 376 patients who had previously gone undiagnosed. This AI intervention led to the identification of new diagnoses for 18 children, including cases of rare neurodevelopmental and neuromuscular diseases. The results highlight a 5% diagnostic yield, which, while seemingly modest, is notable given the extensive prior analyses these cases underwent. This development is groundbreaking for the AI/ML community as it showcases the potential for commercially available AI systems to aid in complex medical diagnostics, expediting the identification of genetic diseases that may have eluded human researchers. The study emphasizes the ability of AI to process sprawling genomic data rapidly, offering new hope for families in search of answers. However, experts caution that while AI has shown promise in medical applications, results still require rigorous human evaluation to ensure trustworthiness. Overall, this research underscores the revolutionary role AI could play in transforming healthcare delivery, particularly in addressing the complexities of rare diseases.