Blog December 16, 2019

New Risk-Based Oversight of Software as a Medical Device

Photo by NEC Corporation of America
Photo by NEC Corporation of America

Digital initiatives and artificial intelligence/machine learning are rapidly becoming central elements of health care. In particular, new tools, apps and devices are fulfilling a variety of needs related to clinical decision support (CDS) that can range from simple tools for maintaining health and wellness to low-risk support tools that help patients and caregivers manage non-serious conditions without a doctor’s assistance. In recognition of the growing importance of such technologies and applications in healthcare — as well as the rapid pace of technological change — the U.S. Food and Drug Administration (FDA) recently published a package of new draft guidance documents that describe how the agency plans to regulate software designed to aid clinical decision making.

The proposed guidelines define software for CDS as technology that can provide doctors, patients, and caregivers with “knowledge and person-specific information, intelligently filtered or presented at appropriate times to enhance health and healthcare.” The definition includes such categories as computerized alerts and reminders for patients and providers, clinical guidelines, condition-specific sets of medical orders, focused data reports and summaries, documentation templates, diagnostic support applications, and contextually relevant reference information. 

The new documents are meant to clarify what software the FDA will regulate as a medical device, and is designed to reflect a flexible, risk-based approach. The agency says it does not intend to regulate low-risk but helpful software, making it clear that software tools such as mobile apps for maintaining or encouraging a healthy lifestyle will fall outside the scope of FDA regulation. Instead, the agency will focus its oversight on higher risk software and machine learning algorithms that are used in serious or critical situations where the program’s logic and inputs may not be fully explained to the user. An example of such higher risk software is an AI-based system that identifies hospitalized patients with type 1 diabetes who may be at risk for cardiovascular events.

The FDA’s changes reflect the need for greater flexibility in the regulation of digital tools and software that can accommodate the rapidly changing technologies and ongoing software revisions needed to keep pace with changes in the marketplace. The agency’s intent is to avoid hindering the development of useful, low-risk health tools, where new versions can be developed much faster than is possible with traditional medical devices.