FDA-cleared artificial intelligence and machine learning-based medical devices and their 510(k) predicate networks

Area of Science:

• Medical Device Regulation
• Artificial Intelligence in Healthcare
• Machine Learning Applications

Background:

• The US Food and Drug Administration (FDA) utilizes the 510(k) pathway for medical device clearance, requiring substantial equivalence to predicate devices.
• There is a growing trend of artificial intelligence and machine learning (AI/ML)-based medical devices being cleared via this pathway.

Purpose of the Study:

• To analyze the predicate networks of AI/ML-based medical devices cleared between 2019 and 2021.
• To investigate the underlying tasks and recall history of these devices.
• To assess the implications for patient safety and regulatory oversight.

Main Methods:

• Network analysis of predicate devices for AI/ML-based medical devices cleared from 2019-2021.
• Examination of device tasks and recall data.
• Comparison of AI/ML device origins and evolution through predicate networks.

Main Results:

• Over one-third of cleared AI/ML devices originated from non-AI/ML predecessors.
• Hematology and radiology devices showed the longest intervals since their last AI/ML predicate (2001).
• Radiology devices exhibited frequent changes in AI/ML tasks across their predicate networks, posing potential safety risks.

Conclusions:

• The current definition of substantial equivalence may not adequately address the unique characteristics of AI/ML in medical devices.
• Increased focus on AI/ML-specific features is crucial for ensuring the safety and efficacy of these evolving technologies.
• Regulatory frameworks need to adapt to the rapid advancements in AI/ML medical devices to safeguard patient care.