Welcome to my Digital Health Regulatory Pathways Master Class series. This series breaks down the core concepts every founder should understand about US and EU regulatory pathways, from what qualifies as a medical device to how risk classification affects your timeline, cost, and market strategy.
You’ve landed on #4 of this series, where I break down all you need to know about risk classes in the US and the EU. It is the most important thing in determining which regulatory pathways are open to you.
Before you start building out documentation or budgeting for regulatory costs, there’s one foundational question every digital health founder needs to answer:
What’s the risk class of your medical device?
Whether you’re pursuing FDA clearance in the U.S. or MDR certification in Europe, this single classification drives nearly every regulatory decision you’ll make.
And if you’re developing a Software as a Medical Device (SaMD) or AI-enabled tool, your classification may not be as obvious as you think.
In this post, we’ll walk you through what “risk class” means in regulatory terms, how the FDA and MDR define and categorize risk, what this classification determines for your startup, and why it’s often more nuanced for SaMD and AI/ML products
Risk class refers to how regulators assess the potential harm your product could cause if it malfunctions or is used incorrectly. It’s not about your product’s quality—it’s about patient safety.
Your risk class determines:
What regulatory pathway you must follow
Whether you’ll need FDA or Notified Body review
What level of technical documentation and clinical evidence is required
How long and expensive your regulatory process will be
In short: no risk class = no roadmap.
The FDA classifies devices into three categories based on intended use, indications for use, and the risk to patients if the device fails.
FDA Class | Description | Examples |
Class I | Low risk | Bandages, exam gloves, tongue depressors |
Class II | Moderate risk | Infusion pumps, pregnancy tests, wheelchairs |
Class III | High risk | Pacemakers, implantable defibrillators |
Find a predicate — Look for a similar product already on the market.
Use a risk-based approach — Evaluate what happens if your product fails.
For SaMD/AI tools: There may not be many good predicates—so expect to rely on the risk-based approach more often than not.
The MDR has 4 risk classes:
MDR Class | Description | Examples |
Class I | Low risk | Stethoscopes, surgical scissors |
Class IIa | Low to moderate risk | Dental fillings, contact lenses |
Class IIb | Moderate to high risk | Ventilators, long-term infusion pumps |
Class III | High risk | Heart valves, implantable neurostimulators |
The MDR uses 23 classification rules (Annex VIII), based on:
Duration of use
Invasiveness
Interaction with the body
Special risk factors (like being an active device or software)
Unlike the FDA, the MDR does not use predicates. You must apply the rules directly—especially tricky for software, which tends to get up-classified even when risk seems low.
Want a deeper dive on that? Check out my post: Where Have All the Risk Class I Medical Devices Gone?
Whether you’re working with the FDA or MDR, SaMD and AI tools often don’t fit neatly into risk categories.
Why? Because these tools:
Are often non-invasive, which makes them look low-risk
Still play a role in diagnosis or treatment decisions
Use logic that may be opaque, dynamic, or difficult to validate
To address this, regulators have issued specific frameworks:
FDA: Software as a Medical Device (SaMD): Clinical Evaluation
EU: MDCG 2019-11 and the IMDRF SaMD risk framework (adopted into MDR)
Key tip: If your SaMD tool supports clinical decision-making, assume it’s moderate or higher risk, especially in the EU.
Determining your risk class early means you can:
Founder’s takeaway: Risk class = time, money, and scope. Treat it as a strategic decision, not just a regulatory label.
In the next post, we’ll turn classification into operational planning and walk you through the documentation required for each risk level, if you need a clinical trial, and how much time and budget should you plan for.
This is where regulatory strategy meets startup execution—don’t miss it.
Really. If you have a question, submit it through this form. I will try my best to respond within a business day. Good questions will get featured on my FAQs list.
Still have questions after exploring? Set up a quick meeting with me here.