AI medical devices: The keys to scaling successfully

Many startups develop high-potential, AI-driven digital medical devices – but few have managed to successfully scale them on the market. One that is making significant strides is RebrAIn, a spin-off of the University Hospital of Bordeaux and the French National Institute for Research in Digital Science and Technology. The company is developing a supervised AI-driven solution that optimises complex, patient-specific neurosurgery. We spoke to its CEO, David Caumartin, when he visited the European Digital Healthtech Conference 2026 in Luxembourg about what digital health entrepreneurs need to keep in mind to succeed.

AI and digital twins in surgery

Where do you see the greatest potential for AI to further transform surgical outcomes?

David Caumartin: I think that the greatest transformative power lies in the development of simple yet efficient AI-powered digital twins – virtual replicas of a patient's anatomy used to simulate and plan surgical procedures. Supervised and explainable AI (models that make their decision-making process understandable, transparent and trustworthy) helps surgeons understand why a certain outcome was achieved for a specific patient and better segment their protocols to optimise results.

There is enormous, untapped potential in the field of complex surgery.
David Caumartin, RebrAIn

The pharma industry was the first to use digital twins to get insights into what types of patients would respond best to different molecules, but there is enormous, untapped potential in the field of complex surgery. AI-enhanced digital twins make it possible to develop quite good models with a fraction of the data needed in the past.

This is particularly interesting in neurology, which is complex, fragmented and, therefore, lacking innovation. The total cost of healthcare for people suffering from brain disorders is rising exponentially, and the use of digital twins can completely transform this field.

Navigating EU and US regulatory approval

What are the key differences between the EU and US regulatory pathways for an AI-powered medical device, and what advice would you give other healthtech companies trying to secure approval on both sides of the Atlantic?

In the EU, we first sought regulatory approval for RebrAIn’s solution under the previous Medical Device Directive, which was replaced by the Medical Device Regulation (MDR) in 2021. This was a simpler procedure, and once done we could commercialise our product and apply for approval from the Food and Drug Administration (FDA) in the US. There as well, with a good advisor, obtaining the necessary certificate was not so complicated.

Receiving MDR approval was considerably more complicated. Fortunately, we had a very good notified body and an auditor who was extremely knowledgeable in AI. A few months after our audit, we received our approval.

The best option is probably to use your home market as your initial pilot and then expand internationally, one market at a time.
David Caumartin, RebrAIn

In the US, our product will probably evolve towards a higher risk class as it becomes more advanced, which might make the next stages of approval more challenging as well. 

What is certain is that right now, it is very hard for a startup to pursue more than two markets simultaneously. The best option is probably to use your home market as your initial pilot and then expand internationally, one market at a time. 

From clinical validation to commercialisation

What has been the hardest part of moving RebrAIn from clinical validation to commercial deployment?

The development of a digital twin model of the brain and the clinical validation phase are both very challenging. The prospective study, including the use of AI in 22 patient cases with submillimetric accuracy and great medical outcomes, was probably our biggest risk so far. Once it was done, we got our first venture capital investment.

Right now, we are engaged in large-scale phase 3 trials including over a hundred patients. To convince neurosurgeons to completely change the way they work, we need compelling data from this randomised, blinded study. Compiling this database is probably the most complicated thing that RebrAIn has ever done.

Scaling an AI healthtech startup

As CEO, what is the single biggest challenge you are grappling with right now, and what would make the biggest difference in overcoming it?

Once you have raised capital, it's a race to reach your biggest milestone with as little time and cash as possible. Right now, for us it is to demonstrate successful results from the phase 3 clinical trial and a certain level of adoption of our tool. Both parts are moving in the right direction: the medical outcomes of the study are great, and RebrAIn’s solution has been tried in 18 different sites, half of which are in the US.

Once you have raised capital, it's a race to reach your biggest milestone with as little time and cash as possible.
David Caumartin, RebrAIn

The next big step will be to go from a handful of paying customers to 30 customers that are paying for, and using, our solution on a recurring basis. When that is the case, we can prove to investors that we have a real, living business model. This is generally the biggest challenge for AI companies.

Advice for digital health entrepreneurs

What would you say is the most important question entrepreneurs in AI and digital health should ask themselves to identify the right way to market access for their device?

The most important thing is understanding how your technology addresses a significant medical need. The medical field is awash with digital innovations, and doctors are constantly invited to try new solutions. If you can’t demonstrate that you have something really big, they will simply not listen.

I chose to focus on AI in neurosurgery because I knew the potential was enormous. Currently, our AI-enhanced neuromodulation therapy (a treatment that uses electrical stimulation to modulate brain activity) targets movement disorders, but it has the potential to address major brain disorders such as epilepsy, depression and addiction. The field is underdeveloped because it is too complex — and that is precisely why using AI in this area is truly captivating.

Doctors are constantly invited to try new solutions. If you can’t demonstrate that you have something really big, they will simply not listen.
David Caumartin, RebrAIn

However, to sell a new solution to neurosurgeons, you need to be really prepared and convincing. Learning how to talk to neurosurgeons in the right way has probably been my biggest challenge during my first two years as CEO of RebrAIn. 

European Digital Healthtech Conference: Key takeaways

Mr Caumartin participated as panellist in the European Digital Healthtech Conference 2026. Organised by Luxinnovation with the support of Medical Valley, the event took place on 26–27 March in Luxembourg and focused on how to achieve genuine adoption of digital medical devices that address real healthcare needs in Europe. Key takeaways from our conversation with Mr Caumartin include:

• AI-powered digital twins can transform complex surgery by enabling patient-specific models with less data
• Healthtech startups should pursue regulatory approval one market at a time
• Phase 3 clinical trials and recurring customers are critical to proving a viable business model
• Addressing a significant, unmet medical need is essential to gain the attention of surgeons
• Learning to communicate effectively with medical professionals is key for healthtech CEOs

As Luxembourg continues to strengthen its position as a hub for digital health innovation, the European Digital Healthtech Conference provides a valuable platform for startups, investors and health sector stakeholders to connect, exchange ideas and explore market access strategies. Pre-register for the 2027 edition to be part of the conversation.