Artificial Intelligence Update: Medical Software and Preemption
In light of the rapidly expanding field of medical software technology, and its recognition that traditional approval mechanisms for hardware-based medical devices may not be well suited to regulating such technology, FDA is piloting a new, streamlined regulatory approach for digital health technologies. The initiative, currently a “working model” and known as the Software Precertification Program, is meant to encourage the development of health and medical software, including potentially software using artificial intelligence.
As currently envisioned, the Precertification Program creates a voluntary, organization-based approach to FDA review of digital health software. FDA will pre-certify organizations as having a “culture of quality” based on FDA’s review of the software developer’s R&D and monitoring systems. Under the working model, pre-certified organizations could submit less information in premarket submissions to FDA than currently required or may qualify for an exemption from premarket review by FDA.
Although it is unknown what specific metrics will be assessed in FDA’s review of organizations in the Precertification Program, the agency has asserted that it will seek to measure Product Quality, Patient Safety, Clinical Responsibility, Cybersecurity Responsibility, and Proactive Culture. Each of these elements will be evaluated in an evidence-based manner and FDA will make a determination regarding certification. Certification status will also be subject to monitoring and revision based on real-world performance data.
FDA’s intent to certify or pre-clear organizations rather than individual products on these safety and effectiveness elements opens a new potential arena for product liability litigation surrounding medical devices. In particular, medical devices are currently governed by an express preemption scheme under which federal law preempts certain state laws that are “different from, or in addition to, any requirement” of federal law. Under that standard, certain lawsuits concerning the safety of a medical device may be preempted, including (1) state-law claims premised on an allegation of fraud on the FDA, and (2) state-law claims involving devices that require pre-market approval, except to the extent those claims simply argue for design or warning requirements that “parallel” federal mandates.
To the extent that lawsuits alleging injury from medical device software (e.g., misdiagnosis) are brought against software developers, resolution of those tort claims will almost invariably involve evaluation by finders of fact of the very elements that FDA intends to examine and pre-certify: whether the software developer has developed, tested, and maintained the software in a fashion that will provide safe and effective patient care. Such suits may, therefore, seek to impose under state law requirements for a particular product that are “different from, or in addition to,” the requirements that FDA has imposed on the development organization as a whole in the pre-certification process. And although courts have not yet considered the applicability of organizational requirements versus product-level requirements in this context, imposing tort liability on software developers who have met FDA’s requirements and are compliant with ongoing oversight programs may disrupt the federal regulatory scheme in the same way that tort lawsuits regarding premarket approved medical devices would. The Supreme Court has previously recognized that such disruption is impermissible.
The outcome of this legal issue will likely depend in part on the methods by which FDA implements the Precertification Program — which are yet to be determined — and on the specificity of its evaluation of individual organizations. Nevertheless, developers should be aware that compliance with the Precertification Program, if and when it is implemented, may have benefits not only in the regulatory setting but also in future litigation down the road.