The American College of Radiology® (ACR®) participated in a U.S. Food and Drug Administration (FDA) public workshop Oct. 14, about “Transparency of Artificial Intelligence/Machine Learning-enabled Medical Devices.” The event provided the regulatory agency with stakeholder perspectives related to data needed by physicians and patients to inform purchase decisions and ultimately safe and effective use of FDA-regulated artificial intelligence (AI)/ machine learning (ML) software.
Keith J. Dreyer, DO, PhD, FACR, Chief Science Officer of the ACR Data Science Institute®, participated in a panel where he highlighted the need for providers to understand which patient groups were represented in AI/ML model training data, how the models were validated and if the models will perform as intended when used with their imaging devices and patient populations. Dreyer and other panel participants noted that different stakeholder groups will have various transparency needs and that providing the necessary data in a usable manner will help pair end-users with the devices that work best for them, ultimately promoting trust and meaningful innovation. He expanded on this presentation in a recent Voice of Radiology blog.
The public workshop was part of the FDA’s extensive efforts to enhance regulatory oversight of AI/ML-enabled medical devices per the January 2021 “AI/ML-Based Software as a Medical Device Action Plan.”
For more information, contact Michael Peters, ACR Government Affairs Director.