Class IIa-certified dental AI: what it means for your clinic
AI is moving fast into dentistry. New tools promise to save time, catch more and support better decisions. But for clinicians, one question matters more than any promise: can you trust it?
We’re proud to share that Umbra, an dental AI model for bitewing X-rays developed and manufactured by Boneprox, is now certified as a Class IIa medical device under the EU Medical Device Regulation (MDR). This makes Boneprox the first Nordic manufacturer to reach this level in its field, and Umbra is now available to clinics through Nanok.
In this article, we explain what the certification means, why it matters when you choose AI for your clinic, and how Umbra works in everyday practice.
What is EU MDR Class IIa?
The EU Medical Device Regulation sets the rules for all medical devices sold in the EU and EEA, including software. Devices are divided into risk classes, from Class I to Class III, and the class determines how strictly a product is reviewed before it can reach clinics.
For software that provides information used in clinical decisions, the starting point is Class IIa. At this level, a manufacturer can’t simply declare its own product compliant. An independent notified body must assess the clinical evidence, the technical documentation and the quality management system behind it. And the review doesn’t stop at certification: the manufacturer remains under ongoing surveillance for as long as the product is on the market.
In other words, Class IIa means the AI has been tested by someone other than the people who built it.
Why certification matters when choosing dental AI
Across Europe, regulators are paying closer attention to AI in healthcare. Under EU MDR, AI software that supports clinical decisions must be certified to at least Class IIa, and the EU AI Act adds further requirements for AI in medical devices.
For clinics, this has a practical consequence. Not every AI tool on the market has been through the same level of review, and it isn’t always easy to tell the difference from the outside. Asking which class a product is certified in, and who assessed it, is one of the simplest ways to know what you’re getting.
We believe clinicians should be able to rely on decision support that has been held to the same standard as any other medical device in the clinic. That’s why we chose the demanding route.
How Umbra works in your clinic
Umbra is built to fit into the way you already work. It analyses bitewing X-rays within your existing workflow and highlights areas with potential findings for you to review.
A few things define how it works in practice:
Fast. Results arrive in under 0.2 seconds (median), so there’s no waiting while the patient is in the chair.
Consistent. The same support in every review, regardless of who is reading the images or how busy the day is.
Clinician-led. The AI identifies. The clinician reviews every finding and makes every decision. Umbra is a decision support tool, not a replacement for clinical judgement.
Backed by published clinical evidence
Certification requires evidence, and Umbra’s is published. In a clinical study published in BMC Oral Health, The model showed 2.2x higher detection performance (mAP) than the top individual expert, and a 60% lower missed-lesion rate (mFNR) than the average expert annotator.
For clinicians, published results mean you don’t have to take our word for it. The evidence is there to read.
Built in the Nordics
Boneprox has worked alongside dentists and radiologists since 2013, building clinical intelligence from inside the clinic. Umbra is the result of that collaboration: developed with clinical input, validated with clinical evidence and certified under Europe’s medical device rules.
Nanok by Boneprox is the team bringing it to clinics, directly and through selected distribution partners across the EU/EEA.
Want to see how AI could work in your clinic?
Get in touch with the Nanok team and we’ll show you how Umbra fits into your workflow, what it takes to get started, and answer any questions you have about AI and regulation.
Comments