Marla Sammer, MD, MHA, Chair of the Pediatric AI Workgroup in the American College of Radiology® Informatics Commission and Vice Chair for Clinical Affairs of the Department of Radiology at Texas Children’s Hospital, contributed this piece.
Artificial intelligence (AI) in medical imaging is rapidly growing to advance healthcare in adults, however, it is far less developed for the pediatric population. Pediatric patient needs must be sufficiently considered during the development, testing and deployment stages of AI to avoid ineffective or detrimental application.
Many of the algorithms commercially developed for use in adults cannot be applied to children due to a variety of factors including a wide range of body sizes, growth and development, different disease types, diverse manifestations of similar disorders, imaging bioeffects and unique socioeconomic factors. Only a handful of tools for image processing or quantification have been developed for use with pediatric patients.
To address the needs of our pediatric patients, the ACR Informatics Commission has established the Pediatric AI Working Group and the Image IntelliGently™ campaign. Image IntelliGently is composed of a wide variety of stakeholders, including private practices, academics and individuals with close ties to other relevant societies including the American Academy of Pediatrics, Healthcare Information and Management Systems Society, Society for Imaging Informatics in Medicine and the Society for Pediatric Radiology, and a variety of expertise in educational, clinical and research domains. Our goal is to ensure equal access to high-quality and safe AI for pediatric patients.
Image IntelliGently supports ACR members’ goals of providing the highest quality of care for all patients, including children. The campaign’s mission is to ensure that AI used in radiology is safe, reliable and effective for pediatric patients. The initial goal is for all radiologists to know that the none of the commercially available AI tools for worklist prioritization are intended to be used on pediatric patients. Practically, this is important since most pediatric imaging isn’t performed in dedicated children’s hospitals. For example, only just over a quarter of pediatric CTs were performed at children’s hospitals, according to a recent publication. Practically, this is important to know since triage AI for adults is being used, but its impact on pediatric patients who are imaged where triage AI is used is not yet known. Consequently, there is reason to be concerned that if radiologists don’t know that AI wasn’t designed for pediatric patients, it may either be used inappropriately, or if it is not used, it may lead to delays in care for pediatric patients that wouldn’t have occurred if triage AI wasn’t being used in adults.
You and your practice have opportunities to get involved by educating and advocating for pediatric AI. In addition, if you’re already using AI with your pediatric patients, we want to hear about your experience. Please visit our resource page for more information.
Please share your thoughts in the comments section below, and join the discussion on Engage (login required).