From hype to reality: What it really takes to enable AI innovation in MedTech

Over the course of an hour, we moved beyond buzzwords to look at the real enablers of AI in MedTech. What we found was clear: the biggest opportunities often come down to getting the basics right.

Here are five key takeaways:

1. AI isn't new – but our perception of it needs an update

As Lavanya Issenmann from Sonova pointed out, AI has been part of our lives for decades. From IBM's Watson on Jeopardy! to Google Maps rerouting our commutes, we use it daily – often without realising. Yet when AI shows up in healthcare, it can still feel unfamiliar, even threatening.

This perception gap creates friction – not just for patients, but also for clinicians introducing AI-powered solutions. As Lavanya put it, the goal isn't just to educate; it's to help people believe. Her team used peer-led testimonials and real-world examples to build trust among audiologists adopting AI-enabled hearing aids. The takeaway? Familiarity is a powerful adoption tool.

2. The biggest barriers aren’t technical – they’re human

Guido Mathews, who leads medical imaging data and AI research at Bayer Radiology, highlighted a revealing paradox – despite hundreds of FDA-approved AI tools, many remain underused. Why? Because hospitals and clinics often lack the infrastructure, integration, and trust to use them confidently.

This isn’t just about regulation or reimbursement. It’s about the day-to-day workflows of radiologists, clinicians, and support teams. If a tool doesn’t fit smoothly into their environment – or if it adds complexity rather than removing it – it won’t get used. Successful adoption depends as much on psychology and process as it does on code.

4. Back to basics: data readiness is non-negotiable

If there’s one phrase that came up repeatedly, it was this: back to basics. As I noted during the talk, scaling AI starts with foundations. And that means data.

Guido stressed that even the most advanced AI solutions struggle to gain traction if the underlying data infrastructure isn’t there. Radiology workflows, for example, are often fragmented – imaging data isn’t always accessible across departments or locations. CDs are still in circulation. Data silos persist. The result? Missed opportunities and wasted potential.

The hard truth is that you can't build scalable, responsible AI on top of broken systems. Data governance, interoperability, and infrastructure investment are essential.

5. What’s next: virtual patients and AI-enabled imaging

Looking ahead, both speakers offered a glimpse into what’s possible when those foundations are in place. Lavanya is especially excited about the rise of virtual patients – synthetic models that could transform diagnostics and personalised care.
Guido, meanwhile, sees huge potential in radiomics and precision imaging. By combining AI with broader patient data sets, we can move beyond 2D tumour tracking and towards holistic, multi-modal insights. It's a shift that could dramatically improve cancer diagnostics and treatment outcomes.

Final thought: from hype to health impact

Innovation in MedTech doesn’t happen through hype. It happens through trust, collaboration, and well-laid foundations. AI has enormous potential to transform healthcare, but only if we’re willing to do the hard work behind the scenes.
At Zühlke, we believe in putting people at the centre of that journey. Because when you get the basics right, the breakthroughs follow.

Curious to learn more or continue the conversation? Watch the full Health Talk here, or meet us at the MedTech Forum 2025.

We’ll be there from May 13–15 in Lisbon, sharing practical insights on AI, data, and real-world adoption. Join our session “From hype to reality: Unlocking AI’s potential in medical devices” on May 14, 12:20–13:10, or stop by booth 10 in the networking area to speak with our experts.