Sidra Medicine helps adapt adult AI tools for newborn care
The study tested whether imaging software trained on adults could spot tube placement problems in newborns while keeping patient data at each hospital.
Qatar's Sidra Medicine has contributed to an international study showing that artificial intelligence models built for adult patients can be retooled for newborn care, the state news agency QNA reported on Wednesday.
The paper appeared in the journal Nature Biomedical Engineering and focuses on improving the accuracy of AI-assisted medical imaging for infants.
The study, titled "Repurposing Adult AI Radiograph Models for Neonatal Care Through Continual Learning: An International Multicenter Study," drew on 30 neonatal intensive care units across 20 countries and territories.
It was carried out through the Medical AI Data for All initiative at Harvard Medical School and examined how existing models could meet the needs of newborns while protecting patient data privacy.
The research looked at endotracheal tubes, which support breathing in critically ill newborns, because incorrect placement can lead to serious complications.
Researchers tested whether AI models trained to check tube placement in adults could identify anatomical landmarks and flag possible misplacement in babies.
The team used a method called continual learning, training an existing model in sequence on neonatal chest X-rays from participating hospitals, with each institution providing 100 to 300 radiographs.
This approach allowed the model to learn from varied patient groups without pooling medical images in one central location.
Sidra Medicine supplied 100 anonymized neonatal chest X-ray images for the work.
Khalid Al-Kharaz, a senior attending physician and co-author of the study, said, "Newborns have unique anatomical and clinical needs."
Al-Kharaz added that the adapted models could help clinicians assess images more accurately and spot problems that need prompt attention.
Another co-author, who manages imaging information systems at Sidra Medicine, said the hospital's participation reflects its commitment to medical innovation through collaboration with leading institutions worldwide.
Abdusamea Shabani, chair of radiology at Sidra Medicine, who selected the radiographs for the study, said continual learning could offer a practical way to adapt imaging models to pediatric settings.
