\u201cThe current lack of labeling standards for AI- or machine learning-based medical devices is an obstacle to transparency in that it prevents users from receiving essential information about the devices and their safe use, such as the race, ethnicity and gender breakdowns of the training data that was used,\u201d she said. \u201cOne potential remedy is that the FDA can learn a valuable lesson from food nutrition labeling and apply it to the development of labeling standards for medical devices augmented by AI.\u201d<\/p>\n
The push for increased transparency around AI-based medical devices is complicated not only by different regulatory issues surrounding AI but also by what constitutes a medical device in the eyes of the U.S. government.<\/p>\n
If something is considered a medical device, \u201cthen the FDA has the power to regulate that tool,\u201d Gerke said.<\/p>\n
\u201cThe FDA has the authority from Congress to regulate medical products such as drugs, biologics and medical devices,\u201d she said. \u201cWith some exceptions, a product powered by AI or machine learning and intended for use in the diagnosis of disease \u2014 or in the cure, mitigation, treatment or prevention of disease \u2014 is classified as a medical device under the Federal Food, Drug, and Cosmetic Act. That way, the FDA can assess the safety and effectiveness of the device.\u201d<\/p>\n
If you tested a drug in a clinical trial, \u201cyou would have a high degree of confidence that it is safe and effective,\u201d she said.<\/p>\n
\n\n\u201cThe current lack of labeling standards for AI- or machine learning-based medical devices is an obstacle to transparency in that it prevents users from receiving essential information about the devices and their safe use, such as the race, ethnicity and gender breakdowns of the training data that was used,\u201d Gerke said. \u201cOne potential remedy is that the FDA can learn a valuable lesson from food nutrition labeling and apply it to the development of labeling standards for medical devices augmented by AI.\u201d<\/p>\n<\/blockquote>\n<\/blockquote>\n
But there are almost no clinical trials for AI tools in the U.S., Gerke noted.<\/p>\n
\u201cMany AI-powered medical devices are based on deep learning, a subset of machine learning, and are essentially \u2018black boxes.\u2019 Their reasoning why the tool made a particular recommendation, prediction or decision is hard, if not impossible, for humans to understand,\u201d she said. \u201cThe algorithms can be adaptive if they are not locked and can thus be much more unpredictable in practice than a drug that\u2019s been put through rigorous tests and clinical trials.\u201d<\/p>\n
It\u2019s also difficult to assess a new technology\u2019s reliability and efficacy once it\u2019s been implemented in a hospital, Gerke said.<\/p>\n
\u201cNormally, you would need to revalidate the tool before deploying it in a hospital because it also depends on the patient population and other factors. So it\u2019s much more complex than just plugging it in and using it on patients,\u201d she said.<\/p>\n
Although the FDA has yet to permit the marketing of a generative AI model that\u2019s similar to ChatGPT, it\u2019s almost certain that such a device will eventually be released, and there will need to be disclosures to both health care practitioners and patients that such outputs are AI-generated, said Gerke, also a professor at the European Union Center<\/a> at Illinois.<\/p>\n
\u201cIt needs to be clear to practitioners and patients that the results generated from these devices were AI-generated simply because we\u2019re still in the infancy stage of the technology, and it\u2019s well-documented that large language models occasionally \u2018hallucinate\u2019 and give users false information,\u201d she said.<\/p>\n
According to Gerke, the big takeaway of the paper is that it\u2019s the first to argue that there is a need not only for regulators like the FDA to develop \u201cAI Facts labels,\u201d but also for a \u201cfront-of-package\u201d AI labeling system.<\/p>\n
\u201cThe use of front-of-package AI labels as a complement to AI Facts labels can further users\u2019 literacy by providing at-a-glance, easy-to-understand information about the medical device and enable them to make better-informed decisions about its use,\u201d she said.<\/p>\n
In particular, Gerke argues for two AI Facts labels \u2014 one primarily addressed to health care practitioners, and one geared to consumers.<\/p>\n
\u201cTo summarize, a comprehensive labeling framework for AI-powered medical devices should consist of four components: two AI Facts labels, one front-of-package AI labeling system, the use of modern technology like a smartphone app and additional labeling,\u201d she said. \u201cSuch a framework includes things from as simple as a \u2018trustworthy AI\u2019 symbol to instructions for use, fact sheets for patients and labeling for AI-generated content. All of which will enhance user literacy about the benefits and pitfalls of the AI, in much the same way that food labeling provides information to consumers about the nutritional content of their food.\u201d<\/p>\n
The paper\u2019s recommendations aren\u2019t exhaustive but should help regulators start to think about \u201cthe challenging but necessary task\u201d of developing labeling standards for AI-powered medical devices, Gerke said.<\/p>\n
![\"\"](\"https:\/\/www.europesays.com\/us\/wp-content\/uploads\/2025\/07\/Gerke-Sara-250331-fz-002-1140x760.jpg\")
<\/a>\u201cThe use of front-of-package AI labels as a complement to AI Facts labels can further users\u2019 literacy by providing at-a-glance, easy-to-understand information about the medical device and enable them to make better-informed decisions about its use,\u201d said Sara Gerke, the Richard W. & Marie L. Corman Scholar at the\u00a0College of Law. Photo by Fred Zwicky<\/p>\n