On Jan. 12, 2021, the Food and Drug Administration (FDA) Digital Health Center of Excellence released its “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan.” The non-binding policy document provides a high-level overview of five actions the agency intends to take over the next year, including:
- Development of draft guidance on “predetermined change control plans” to expedite developer-planned future modifications to AI/ML-enabled software.
- Continued efforts to collaborate with industry and AI/ML experts on “Good Machine Learning Practice (GMLP).”
- Promotion of AI/ML transparency and trustworthiness for end-users and patients.
- Identification and elimination of bias in algorithms and improvement of algorithm resilience to changing inputs and clinical context.
- Engagement with stakeholders to pilot real-world performance monitoring.
The action plan was informed by several key FDA activities from the past two years, including public comments received on its “Proposed Regulatory Framework for Modifications to AI/ML-Based Software as a Medical Device,” the February 2020 public workshop on the “Evolving Role of AI in Radiological Imaging” and the October 2020 meeting of the FDA Patient Engagement Advisory Committee on enhancing patient trust in AI. The American College of Radiology® (ACR®) participated in each of these FDA activities under the leadership and guidance of the ACR Data Science Institute® and Commission on Informatics.
- ACR comments on FDA’s proposed framework for modifications to AI/ML (June 3, 2019).
- ACR Leaders Participate in FDA Workshop on AI in Radiological Imaging (Feb. 27, 2020).
- ACR and RSNA Highlight Challenges to Future FDA Oversight of Autonomous AI (July 1, 2020).
- ACR DSI Presents to FDA Patient Engagement Advisory Committee on AI (Oct. 22, 2020).