Privacy-preserving training could expand medical AI for children while keeping data local
In the field of radiology, AI models have the potential to improve imaging analysis, speed up diagnoses and catch errors that might otherwise be missed. But the vast majority of these models are trained on adult data and don't translate well to infants and children, keeping these advances out of reach of neonatal and pediatric care.
Radiology AI models show promise in improving diagnosis and reducing errors, but most focus on adult data and struggle to apply to pediatric patients. Vishwanatha Rao from Harvard Medical School explains that children's medical records are tightly regulated, often remaining within individual institutions, limiting data access for developing robust AI models.
A study published in Nature Biomedical Engineering reveals that adult-trained models can be adapted for infants by using data from multiple hospitals, without requiring sensitive information to leave each institution. The team trained a model on 2,965 chest X-rays from 30 neonatal intensive care units across 20 countries, adapting it to identify endotracheal tube placement with an average error of 0.67 cm, significantly better than single-site adaptations and adult models.
This privacy-preserving method could extend to other adult models for pediatric use, addressing gaps in pediatric medical AI deployment. The researchers believe this approach could broaden AI applications to underrepresented patient populations.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
