Doha, November 12, 2026 : Sidra Medicine, a member of Qatar Foundation, has contributed to an international study exploring how artificial intelligence (AI) models initially developed for adult patients can be adapted to improve medical imaging for newborns.
Published in the journal Nature Biomedical Engineering, the research demonstrates a potential approach to making AI-assisted imaging more accurate for neonatal care while safeguarding patient data privacy.
Titled "Repurposing Adult AI Radiograph Models for Neonatal Care Through Continual Learning: An International Multicenter Study," the research involved 30 neonatal intensive care units across 20 countries and territories. It was conducted through the Medical AI Data for All (MAIDA) initiative at Harvard Medical School.
The study examined how existing AI systems could be modified to address the distinct anatomical and clinical requirements of newborn patients.
Dr. Khalid Al-Kharaz, Senior Attending Physician and Director of the Pediatric Neurosurgery Fellowship Program at Sidra Medicine, and a co-author of the study, said AI could play an important role in supporting clinical decisions, provided that its performance is reliable across different patient groups.
He noted that newborns have specific medical and anatomical needs that technologies designed for adults may not adequately address. The research demonstrates how existing AI models could be adapted to help clinicians assess medical images more accurately and identify conditions requiring timely attention.
Dr. Khalid Al-Kharaz, identified in the study announcement as Manager of Imaging Information Systems at Sidra Medicine and a co-author, said the institution's involvement reflects its commitment to medical innovation through international research partnerships. He added that contributing clinical imaging expertise and data supports efforts to address gaps in paediatric healthcare and develop AI technologies that better meet children's needs.
Dr. Abdusamea Shabani, Chair of Radiology at Sidra Medicine, who selected the radiographs used in the research, described the study as an important step towards addressing the limitations of AI systems developed mainly for adult patients. He said continual learning could offer a practical way to adapt existing imaging models for paediatric care while protecting patient privacy.
The research focused on checking the positioning of endotracheal tubes, which help critically ill newborns breathe. Correct placement is essential in neonatal intensive care, as incorrect positioning can result in serious complications.
Researchers examined whether AI models trained to assess tube placement in adults could be adapted to identify anatomical landmarks and detect potentially incorrect positioning in newborns.
To achieve this, the team used a method known as continual learning, which involves updating an existing AI model through successive training stages. Neonatal chest X-rays from participating hospitals were used for this process, with each institution contributing between 100 and 300 images.
This approach enabled the model to learn from varied patient populations without requiring medical imaging data to be centrally pooled. Sidra Medicine contributed 100 anonymised neonatal chest X-ray images to the study.
Source: QNA
