Marla Sammer, MD, MHA, Chair of the Pediatric AI Workgroup in the American College of Radiology® (ACR®) Informatics Commission and Vice Chair for Clinical Affairs of the Department of Radiology at Texas Children’s Hospital, contributed this piece.

Pediatric patients are confronted with significant healthcare disparities: primarily a shortage of specialized radiologists and the fact that most pediatric care in radiology is provided in adult-focused settings. Artificial Intelligence (AI) could be a robust tool to bridge these disparities, but it's essential that it is developed with pediatric patients in mind, and to use it responsibly to prevent unintended harm.

AI holds promise for pediatric radiology. For example, if more tools are developed, AI could help address the severe shortage of specialized pediatric radiologists by enhancing the skills of radiologists who have not specialized in pediatrics. These benefits are in addition to those offered by AI for patients in radiology, but to date, largely have not made their way to pediatric patients. These benefits include the potential to speed up patient care, increase radiologists’ efficiency by automating time-consuming tasks, revolutionize care for children with rare diseases and provide predictive analytics to extract additional insights from radiology data and generate clinically useful patient specific insights such as likelihood of disease progression, opportunistic screening, and early disease detection for children.

Though the use of AI has rapidly expanded in recent years, very little AI has been dedicated to pediatric imaging, likely due to several interconnected factors. This includes a lack of child-specific datasets, a shortage of pediatric radiology subspecialists to help with development, undefined AI reimbursement structures, the overall smaller market share and stricter research regulations. Additionally, pediatric patients' inherent diversity in diseases, size, physiology and anatomy make it difficult for a single AI solution to fit all pediatric patients.

Solutions to these issues involve comprehensive training and validation protocols for AI. Additionally, creating a structure for AI reimbursement that incentivizes inclusive patient care, developing a more collaborative research and development infrastructure, and optimizing regulatory oversight could pave the way to more AI development for pediatric patients.

The potential of AI to improve pediatric radiology is within reach. By addressing these issues, we can enable a future where all children can receive safe, effective, reliable and equitable access to the highest quality of care in radiology.

Finally, on behalf of the ACR Pediatric AI Workgroup, please read our new white paper in the Journal of the American College of Radiology. It discusses the health equity issue of the lack of pediatric AI and offers solutions to improve pediatric AI. Together, we can enable a future where all children receive fair and equitable care.

Please share your thoughts in the comments section below and join the discussion on Engage (login required).

×Your comment has been successfully submitted for approval and will be published after approval.
Leave a Comment

You may also like