Building the Foundations for Trustworthy AI in Pediatrics.

Artificial intelligence (AI) adoption in health care is rapidly accelerating, with pediatrics positioned to benefit from improvements in quality, efficiency, and access to care. Recent frameworks have articulated ethical commitments for trustworthy AI in pediatrics, but less attention has been paid to the foundational infrastructure required to achieve them. In this article, we examine the challenges that emerge at the intersection of AI and pediatrics: unique challenges that AI poses for pediatric care, including nonintuitive errors, impacts on clinician cognition, and new infrastructure requirements; and unique challenges that pediatrics poses for AI, including adequate training data, accommodation of developmental heterogeneity, and navigation of evolving patient autonomy. We then outline the data systems, governance structures, validation frameworks, and public trust infrastructure that must be established before the field can meet these challenges. We conclude with concrete recommendations for clinicians, caregivers, health care systems, government, technologists, and academic institutions seeking to lay the foundation for trustworthy and beneficial AI in pediatrics.