Predetermined Change Control Plan Adoption and Documentation Transparency in FDA-cleared Radiology Artificial...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Regulatory Science

Background:

• The U.S. Food and Drug Administration (FDA) established Predetermined Change Control Plans (PCCPs) to manage modifications in artificial intelligence/machine learning (AI/ML) devices.
• Transparency in documentation for AI/ML devices is crucial for postmarket surveillance and patient safety.

Purpose of the Study:

• To assess the adoption rate and documentation completeness of PCCPs for FDA-cleared radiology AI/ML devices.
• To evaluate the transparency of PCCP documentation concerning device lifecycle management.

Main Methods:

• A systematic scoping review of FDA AI/ML-enabled device databases (2015-2025) was performed.
• PCCP documentation completeness was scored using an 8-point rubric based on FDA guidance.
• Manual verification of 34 radiology AI/ML devices with PCCPs was conducted against FDA regulatory summaries.

Main Results:

• Radiology AI/ML devices showed high PCCP adoption (91.9%) after the final FDA guidance in December 2024.
• PCCP documentation scores averaged 5 out of 8, with common modifications in retraining and optimization.
• Public summaries often omitted continuous performance monitoring and predefined drift triggers for retraining.

Conclusions:

• PCCP adoption in radiology AI/ML devices surged post-guidance, but transparency regarding lifecycle controls remains limited.
• Standardized reporting of lifecycle controls in PCCPs is recommended for effective postmarket monitoring.
• Enhancing public documentation of PCCPs is essential for systematic oversight as AI/ML device pathways expand.