Classification of Medical Devices: Why It Matters, More Than Ever
Recently, a lively discussion unfolded on our social media channels about how a particular medical device should be classified. This is not a trivial question. Classification determines how a device reaches patients, who can buy it, whether medical supervision is required, and even whether surgery is needed to implant it.
Although regulatory frameworks differ across continents, one principle remains universal: the higher the risk, the tighter the regulation. Safety is the cornerstone.
But here is where things get interesting: technology is evolving far faster than regulation. Artificial intelligence, autonomous systems, and next-generation robotics are challenging the foundations of how we define a “medical device.” And some of the decisions regulators made years ago may no longer make sense today.
Let’s break this down.
How Are Medical Devices Classified—and Why Should We Care?
The United States: FDA Classification
The FDA classifies devices into three categories based on the risk they pose and the level of regulatory control required to ensure safety and effectiveness:
- Class I Devices not intended for use in supporting or sustaining life or of substantial importance in preventing impairment to human health, and which do not present a potential unreasonable risk of illness or injury.
- Class II Devices for which general controls alone are insufficient to provide safety and effectiveness.
- Class III Devices that support or sustain life, are implanted, or present a potential unreasonable risk of illness or injury.
In short: Class I → low risk, Class II → moderate risk, Class III → high risk.
The European Union (EU): CE Marking & MDR
In Europe, every product intended for a medical purpose must meet the requirements of the Medical Device Regulation, MDR (EU 2017/745) and obtain CE (Conformité Européenne) marking.
The EU also classifies devices based on risk and invasiveness. And this is where things get very nuanced, especially for devices that enter the body through natural orifices, are used for short periods, or interact with active systems.
To obtain CE marking, a device must undergo a conformity assessment, which becomes progressively more demanding as the device’s risk increases. The EU uses four classes:
- Class I — low risk. Examples: bandages, non-sterile surgical instruments, simple dental devices. Manufacturers can often self-certify compliance, except for special subsets (Is, Im, Ir).
- Class IIa — medium risk (lower). Examples: dental fillers, hearing aids, diagnostic ultrasound probes. Requires assessment by a notified body.
- Class IIb — medium–high risk. Examples: infusion pumps, ventilators, long-term invasive devices. Significant regulatory scrutiny.
- Class III — high risk Examples: implantable pacemakers, heart valves, brain stimulators. Requires the strictest assessment, including clinical investigations.
Classification is determined by 22 rules in Annex VIII of MDR, grouped into four major categories:
- Non-invasive devices
- Invasive devices
- Active devices (including software and robotic systems)
- Special rules (sterility, biological materials, nanomaterials, etc.)
Invasiveness, duration of use, anatomical location, and interaction with active systems are key determinants.
As a general rule:
- More invasive = higher class
- Longer duration = higher class
- Active or energy-delivering systems = higher class
- Life-sustaining or implantable = Class III
More importantly, European regulation is currently in a state of transformation. Emerging rules now address cybersecurity, connectivity, and, crucially, the use of artificial intelligence in medical devices or Robotic Assistance and autonomy. This reality forces us to confront new ethical questions and new safety paradigms, questions the current classification systems were never designed to answer. These issues were barely imaginable when the old MDR rules were created. Now, they are redefining what a “medical device” even is.
Two Examples to Understand the Problem
Mandibular Advancement Devices (MAD)
In a recent post we analysed a particular Mandibular Advancement Device. According to a European notified body evaluating the device under MDR 2017/745 It is an invasive, non-surgical device because it enters the oral cavity. It is used short-term, approximately 8 hours per day. According to Rule 5, paragraph 2: Short-term invasive devices used in the oral cavity typically fall under Class I.
Conclusion: this MAD is a Class I medical device in the European Union.
This means it faces the lowest level of regulatory scrutiny, even though it modifies airway anatomy during sleep. It’s worth pausing a moment to reflect on that, because most of the professionals involved in the care of Sleep Apnea patients would say it clearly is a Class II Medical Device.
Surgical Robots—Yes, Really
Here comes the part that surprises many clinicians.
The da Vinci Surgical System, the most widely used surgical robot in the world, is classified by the FDA as a Class II device. How is this possible?
Because it was cleared through the 510(k) pathway, a mechanism originally designed in 1976 to allow “non-substantial modifications” of pre-existing devices. This is not unique to soft-tissue robots. Systems like ROSA, MAKO, and ROBODOC fall under similar pathways as “orthopaedic stereotactic instruments.”
And almost all surgical robots commercially available today in the United States were cleared as not substantial modifications of devices that existed before 1976 (!?).
In Europe, the da Vinci Xi obtained CE marking as a Class IIb device.
Given everything above, let’s ask the question that every surgeon naturally asks: If you had to classify a surgical robot today—without considering historical pathways or legacy loopholes—what class would you put it in?
A device that:
- Enters the body,
- Holds instruments inside the patient,
- Performs surgical actions,
- And may progressively incorporate AI-driven assistance…
Wouldn’t almost every surgeon instinctively classify it as Class III? Should systems that we use to perform surgical actions be regulated as moderately risky devices? Are we comfortable with this? Should we rethink the entire classification system now that robotics and AI are rewriting the fundamentals of intervention? These are the questions we should be discussing—urgently.
Final Thought
Medical device classification is not a bureaucratic exercise. It is a patient safety issue, a clinical practice issue, and increasingly, a societal issue.
As robotics and AI become central to modern medicine, our regulatory frameworks must keep pace. Otherwise, we risk navigating 21st-century technologies with 20th-century rules.
What do you think? Join the debate. Share your view.
J Granell, Dec 7, 2025
Postdata. As I finish writing this post, I can hear my wife in the next room passionately debating on the phone a completely different, yet strangely parallel, issue: the growing concern that modified human organs might be classified as medicines rather than as human organs for transplantation. Such a shift would place them under an entirely different regulatory ecosystem, potentially opening the door to commercialization and profit, and breaking a basic ethical rule that governs human organs donation and transplantation worldwide: nonprofit. There are already movements within the European Community to prevent this. It is a powerful reminder that due to the fast-changing world of medical devices, tissues&organs, and biotechnology in general, the borders between are no longer clear. And as these borders blur, the rules that govern them become battlegrounds for ethics, safety, and economic interests. And the debate is not abstract. It’s happening right now.