HIMSS, PCHAlliance Comment on FDA Proposed Regulatory Framework for Changes to Artificial Intelligence/Machine Learning

HIMSS, PCHAlliance Comment on FDA Proposed Regulatory Framework for Changes to Artificial Intelligence/Machine Learning

On June 3, HIMSS and PCHAlliance submitted a joint comment to the Food and Drug Administration (FDA) in response to its recent discussion paper and request for feedback on a Proposed Regulatory Framework for Modifications to Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD). With this proposed framework, FDA envisions that “with appropriately tailored regulatory oversight, AI/ML-based SaMD will deliver safe and effective software functionality that improves the quality of care that patients receive.”

HIMSS and PCHAlliance both found this proposal to be a positive step in the advancement of medical device technology because it proactively addresses a regulatory gap for advanced AI/ML that has the potential to have a distinct impact in healthcare. One issue that the paper makes clear is the fact that the current regulatory paradigm was not designed to account for the rapid adaptation potential of AI/ML technologies and thus, this update is both timely and desirable to keep pace with the rate of change in technology, while ensuring that quality assurances are maintained by the software developer community.

Despite the overall positive outlook of this proposal, HIMSS and PCHAlliance raised concerns with regard to how the interdependence of this proposed framework and FDA’s Software Pre-Certification Pilot program make the success of the proposed framework contingent on the final version and success of the Pre-Cert program. Recommendations were made for greater focus and clarity on elements related to the types of changes that require FDA review and oversight, as well as the methodologies for verifying and validating adaptive AI/ML algorithms in a manner that supports assurance of reasonably safe and effective algorithms in the market. Moreover, HIMSS and PCHAlliance laid out security and privacy concerns related to data usage for the purposes of creating test training sets, noting that existing vulnerabilities may lead products to be susceptible to malicious use of AI/ML.

HIMSS and PCHAlliance was pleased to see the lengths the proposal goes to when describing the necessity of quality systems in place, manufacture’s practices, and training sets within the proposed Total Product Lifecycle Approach. However, it was noted in the response that while the potential is appreciated, the Total Product Lifecycle Approach is conceptual at this point and has not yet been proven to enable the goals identified.

Going forward, HIMSS and PCHAlliance recommended FDA convene a technical expert panel and/or leverage the expertise from participants currently involved in the Pre-Cert program to advise future frameworks such as the one in this proposal. Another recommended alternative was for FDA to consider establishing an ongoing program such as “Smart and Connected Health” that focuses on the interfaces between research and industry domains to identify and maintain key issues and share learnings.

View the full letter

Published on