Is my product a medical device? EU Court judgment clarifies that the test includes objective function and intended purpose, and provides guidance relevant to software and AI-enabled tools

The recent judgment of the Court of Justice of the European Union (CJEU or the Court) in Case C-427/24, Diagramm Halbach clarifies the key principles for determining whether a product falls within the definition of a medical device and will be subject to the EU medical device regime. Crucially, the CJEU held that this analysis includes an assessment of both the manufacturer’s intended purpose of the product and whether the product objectively performs functions for a specific medical purpose listed in Article 2(1) of Regulation (EU) 2017/745, the Medical Devices Regulation (MDR).

The judgment also provides helpful guidance for borderline products, including certain types of software and AI-enabled tools, for which the application of the classification rules is still developing, building on the Court’s prior judgment in Snitem (C-329/16).

In Case C-427/24, Diagramm Halbach, the Court considered both whether patient identification wristbands were medical devices and what role the manufacturer’s “intended purpose” plays in this assessment. German authorities had alleged that the wristbands were medical devices due to the way they were advertised, whereas the manufacturer emphasized that the wristbands — which were supplied “blank: — served a purely administrative function to identify patients. The judgment helpfully outlines the test for “intended purpose,” including in the context of software used in a healthcare context.

Intended purpose is to be inferred from the actual use intended by the manufacturer and objective functionality

The MDR definition of a medical device starts with its “intended purpose.” Article 2(12) MDR defines “intended purpose” as the use for which a device is intended according to data supplied by the manufacturer on the label, in instructions for use, promotional or sales materials or statements, and as specified by the manufacturer in the clinical evaluation.

In accordance with this definition, the Court acknowledged that promotional or sales materials can contribute to the assessment of intended purpose, as the Court previously held in Snitem (C-329/16).

However, the Court has now clarified that the role played by promotional and sales material in determining the intended purpose has limits. Specifically, the Court emphasized that not every statement in such material is decisive; the relevant question is whether the data actually relates to the use of the product, rather than merely providing contextual information designed to promote adoption.

In the present case, the manufacturer’s promotional claims highlighted the importance of high‑quality patient identification wristbands and claimed that medication errors were the third-leading cause of death. However, the Court considered that such information did not provide any indication as to the medical use of the patient identification wristbands — the use of the wristbands had been solely described as patient identification.

The Court also confirmed that the manufacturer’s presentation of the product is not conclusive on its own; the product must also objectively perform one of the functions for the specified medical purposes listed in Article 2(1) MDR. The wristbands provided no such function, as they merely identified the patient; therefore, this criterion for qualification as a medical device was not met.

Distinction between the administrative identification of patients from diagnosis, monitoring, or treatment and other medical purposes

The Court’s reasoning drew a clear line between medical purposes on the one hand and administrative or organizational support on the other. By way of illustration, the method used to identify a patient has no direct effect on medication administration, procedures, transfusions, or sample collection; wristbands facilitate identification by healthcare staff, but so would wearing a nameplate or carrying a copy of an identification document. The wristband does not itself perform a diagnostic or therapeutic function.

This is the key practical point with implications beyond the current case: a product can be important to a healthcare workflow and the outcome of using the product could reduce patient harm, without the product necessarily becoming a medical device. Marketing claims that refer to potential risks and patients’ safety will be relevant to the classification analysis, but they do not displace the need for the product to have a medical function to be a medical device. MDR medical device classification rules and Medical Device Coordination Group (MDCG) guidance continue to be central when considering medical device qualification and classification.

Configurable products

One point of interest is that the wristbands in question were supplied “blank” (i.e., without any printed patient details), which supported the Court’s conclusion that the products were not medical devices. In this regard, the Court stated that “it must be borne in mind that those identification wristbands were supplied unprinted. It follows that they are not intended to serve a specific medical purpose or to exert a physical action on the human body.” While this implies the possibility that the wristbands could be medical devices if they contained certain printed information, the Court did not go on to explain how any printed information such as the first name, last name, date of birth, and file number of the patient and hospital department could justify such a classification (in fact, the Court noted printed medical information would perform the same function). At any rate, it is notable that the mere possibility of such subsequent printing and customization did not affect the Court’s classification at the point of supply. There are limited lessons, therefore, from this judgment in terms of downstream customization of a healthcare product.

What does this mean for digital health and AI-enabled tools?

The judgment is particularly relevant to EU digital health technologies, including AI-enabled ambient scribe tools, workflow software, and clinical decision support systems: a product used in a medical context is not automatically a medical device; rather, it must be intended for a specific medical purpose and be capable of performing a medical function.

Administrative software that stores, archives, transmits, searches, or formats information may support clinical workflows without necessarily performing a medical function. For example, an AI scribe that transcribes a consultation, drafts a referral letter, structures notes, or populates an electronic record may be positioned as an administrative tool, depending on its claims, outputs, intended use, and actual function.

The analysis changes where software moves beyond performing administrative tasks. A tool is more likely to fall within the medical device definition where it identifies a condition, predicts deterioration, calculates risks, ranks differential diagnoses, recommends treatment, suggests dosing, or otherwise generates outputs intended to inform diagnosis, prevention, monitoring, prediction, prognosis, or treatment decisions.

For AI-enabled products, the assessment should look at individual functions as well as the whole product; a single platform may perform both administrative functions and medical-device functions and, in some cases, it may make sense to separate these into different modules with different classifications. Companies should map features one by one, test claims against functionality, and consider whether each output is intended merely to support workflow or to influence clinical judgment.

Practical Takeaways

• Manufacturers should apply the two-part test: Consider both the intended purpose and whether the product objectively performs a function for a specific medical purpose under Article 2(1) MDR. Contextual information, which is merely useful in nature, such as that contained in advertising, is not decisive.
• Manufacturers should focus on function, not just the setting: Use by clinicians, deployment in hospitals, or contribution to safer care does not necessarily make a product a medical device. The key questions are what the product’s output is intended to do and does it perform a medical function.
• Assess software and AI feature by feature: Distinguish administrative functions such as recording, formatting, or transmitting information from functions that influence diagnosis, monitoring, prognosis, or treatment.
• Consider configurable products carefully: Downstream customization does not automatically determine classification, but a blank or modular product may still be a medical device where the manufacturer designs or promotes it for a medical purpose and it could fulfill that function.
• Document and revisit the analysis: Keep intended-purpose statements, claims, specifications, and product functionality aligned, and reassess classification when features, outputs, integrations, use cases, or marketing materials change.

Trainee solicitor Teresa Menendez Moral also contributed to this blog post.

This post is as of the posting date stated above. Sidley Austin LLP assumes no duty to update this post or post about any subsequent developments having a bearing on this post.