OpenAI's o3 model is making waves in the medical world by solving cold cases in rare genetic illnesses.
A groundbreaking study published in NEJM AI reveals how this AI tool has helped Boston Children's Hospital diagnose 18 children with previously unsolved rare diseases. These families, once left in the dark, now have answers, thanks to AI's prowess.
The study, led by Catherine Brownstein, PhD, at the Manton Center for Orphan Disease Research, ran the genomes of 376 undiagnosed patients through the o3 system. Astonishingly, the AI increased new diagnoses by 5%, a significant achievement considering the complexity of these cases. Brownstein describes this as a "total game changer," highlighting the profound impact on families seeking answers.
The AI model analysed clinicians' notes, patient symptoms, and potential gene lists. Its outputs were meticulously reviewed using standard criteria before confirming any diagnosis. The research uncovered new diagnoses across four disease areas, including neurodevelopmental and neuromuscular disorders.
Brownstein emphasises that AI is not a replacement for human expertise but a tool to alleviate bottlenecks in genetic diagnosis. AI can sift through vast data without fatigue, allowing geneticists to focus on promising leads. However, caution is necessary to avoid over-reliance on AI, as it can produce coherent yet incorrect explanations, known as hallucinations.
Patient privacy was paramount, with data minimisation ensuring no protected health information was used. Brownstein envisions a future where AI democratises access to genetic analysis, aiming for a publicly accessible reanalysis tool within a few years. This AI-driven approach promises to revolutionise medical diagnostics, offering hope to countless families.
Want to hear more? Join Mal & Matt on the Property AI Report Podcast each week!
Access from your preferred podcast provider by clicking here
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai