In our last post, we took a brief look back through history at FDA’s approach to regulating medical device software and found that there is little distinction from the agency’s approach to hardware devices. Recently, however, FDA has announced several digital health initiatives aimed at improving the agency’s resources and policies governing software and data systems (including its own internal data systems) and changing the way the agency handles pre-market reviews of and compliance activities for software as a medical device (SaMD) and SaMD manufacturers. In this post, we will review FDA’s digital health improvement highlights from the past few years and take a quick look at the agenda for the transparency of AI/ML-enabled medical devices workshop scheduled for October 14, 2021.
Some of FDA’s most recent initiatives, and those most relevant to the upcoming workshop, relate to AI/ML-based SaMD regulation. In particular, on April 2, 2019, FDA published a discussion paper titled Proposed Regulatory Framework for Modifications to Artificial Intelligence/Machine Learning-Based Software as a Medical Device, which described the application of the International Medical Device Regulators Forum (IMDRF) risk-based framework to AI/ML-based SaMD and introduced possible lifecycle controls for such devices, such as good machine learning practices (GMLP), SaMD pre-specifications (SPS), algorithm change protocols (ACP), and real-world performance monitoring. The discussion paper posed multiple questions to gauge stakeholder reactions to the proposed framework, including transparency and real-world performance monitoring of AI/ML-based SaMD. In October 2020, the Agency held a Patient Engagement Advisory Committee (PEAC) meeting devoted to AI/ML-based devices in order to gain insight from patients into what factors impact their trust in these technologies.
Then, on January 12, 2021, FDA released the Artificial Intelligence and Machine Learning (AI/ML) Software as a Medical Device Action Plan (Action Plan), which responded to stakeholder comments on the discussion paper and described the agency’s planned next steps towards developing greater oversight of AI/ML-based SaMD. The action plan expressly includes the upcoming workshop on AI/ML-based SaMD transparency as a next step to advance a patient-centered approach to the development and regulation of such devices. According to the Action Plan, FDA plans to discuss insights gained from the PEAC meeting and to elicit input from the broader community on how device labeling supports transparency to users at this week’s workshop.
In 2017, FDA announced the launch of its Software Precertification (or Pre-Cert) Pilot Program, kicking off the agency’s formal development and evaluation of a new regulatory approach that the agency believes will promote SaMD innovation while assuring safety and effectiveness. In brief, under Pre-Cert, FDA conducts an Excellence Appraisal to assess a SaMD manufacturer’s organizational practices and quality system to determine whether a manufacturer is eligible for precertification. Once precertification is obtained, the manufacturer is eligible for streamlined review of new products or modifications to existing products and, in some cases, may bypass premarket review altogether.
The promise of the Pre-Cert program is that it will focus regulatory scrutiny on the manufacturer and its practices to ensure product quality rather than examining each new product and modification. This approach would certainly address some of the issues with FDA’s traditional approach to software regulation we described in our previous post.
The Pre-Cert program is still in pilot phase with nine participating manufacturers, and it may be years before SaMD manufacturers are able to apply for Pre-Cert status. FDA’s website on the Pre-Cert program may be found here. In addition, you can find our previous observations on the program here and here.
Digital Health Center of Excellence
In September 2020, FDA launched the Digital Health Center of Excellence (DHCoE) to “provide centralized expertise and serve as a resource for digital health technologies and policy for digital health innovators, the public, and FDA staff.” The DHCoE is currently leading FDA’s regulatory efforts relating to cybersecurity, the Pre-Cert Pilot Program, AI/ML-based SaMD, and wireless medical devices. FDA’s concentration of resources in the DHCoE will likely significantly improve the agency’s regulatory policies and rulemaking relating to new developments in software and digital technology.
You can find FDA’s description of the DHCoE here.
The public workshop will take place on Thursday, October 14, and will begin at 10 AM with opening remarks from Nooshin Kiarashi (Digital Health Engineer, Lead Reviewer, Office of In Vitro Diagnostics and Radiological Health), Jeff Shuren (Director, Center of Devices and Radiological Health), Bakul Patel (Director, Digital Health Center of Excellence), and Matthew Diamond (Chief Medical Officer, Digital Health Center of Excellence). At 10:30 AM, there will be presentations and a panel discussion focusing on two questions:
- “What does transparency mean to different stakeholders of AI/ML-enabled medical devices?”
- “What information about AI/ML-enabled medical devices is important to each stakeholder for establishing the safety and effectiveness of these products?”
After the lunch break, there will be a 50-minute open public comment session, where registrants who have requested an opportunity to speak will be able to present brief comments on the topic of AI/ML-based SaMD transparency. Finally, at 1:30 PM, there will be another round of presentations and panel discussion on the following question:
- “What are the current and possible means of promoting transparency of AI/ML-enabled medical devices to users? What practices are the stakeholders seeing for information sharing, including labeling? What are important areas for future development?”
In this post, we provided a brief overview of FDA’s recent actions to develop new digital health programs and regulatory processes. Next week, we will share our comments and takeaways from the transparency of AI/ML-based medical devices workshop.