Regulatory
rigorously, and governed appropriately?’” Zimmer Biomet says responsible AI should run through the product life cycle rather than sit beside it as a separate exercise. Its portfolio includes OrthoGrid Hip AI, an AI-powered platform for total hip arthroplasty, and ZBEdge Analytics, which connects intra-operative, mobility and outcome data. “The companies leading the next era of medtech
won’t simply be those building the most powerful algorithms. They’ll be the ones earning trust by demonstrating their AI is safe, transparent and responsibly developed at every stage. For us, responsible AI isn’t a separate initiative. It’s a core design principle, [an] essential ingredient in building the next generation of orthopaedic technology,”
Assessing orthopaedics For Dr James Baker, an orthopaedic surgeon at UofL Health who uses Zimmer Biomet platforms and sits on one of its advisory boards, regulation and clinical usability are closely linked. “The question the AI Act asks manufacturers is one surgeons have asked in operating rooms for a while. Not what the system can do – what it changes, and whether anyone can show it.” He argues that connected and robotic systems are already creating an evidence trail that manufacturers must learn to govern. “One case now produces more measurement than a surgeon a generation ago saw in a month of follow-up. The regulatory question is whether that data is governed. The clinical question is whether it reaches the next decision. I’ve stopped treating those as different questions.
“What I’d watch between now and 2027 isn’t whose documentation runs longest. It’s which manufacturers built their systems so human oversight was already happening before anyone required it, where the clinician can see what the algorithm considered. Where the path from output to decision is visible enough to audit because it was designed to be visible enough to use.”
That emphasis on visibility also fits the emerging legal framework. High-risk AI systems under the Act must be designed for effective human oversight and sufficient transparency for deployers to interpret and use outputs appropriately. For qualifying high- risk systems, those duties are concrete rather than conceptual. Providers must address risk management; training; validation and testing data; technical documentation; record-keeping; transparency and human oversight; and accuracy, robustness and cybersecurity. The Act also requires a quality management system and post-market monitoring. EU guidance on the interaction between the AI Act and the MDR/IVDR encourages manufacturers to integrate overlapping testing, reporting and documentation where appropriate.
“A company that involved surgeons early will find the deadline mostly describes what it already does,” Baker continues. “That involvement isn’t a compliance
www.medicaldevice-developments.com
strategy. It’s how a system gets built to work on a real day, in a real hospital, in a working surgeon’s hands. “None of that settles the documentation the Act demands. But usability and compliance tend to fail in the same place, and the systems built to be used are the ones that will have less to explain. The regulation is arriving at the same conclusion the operating room reached first.” An Arthroplasty Today study offers one example of the evidence manufacturers may have to build. Comparing AI-controlled and human-controlled versions of the same fluoroscopic navigation platform in 420 total hip arthroplasties, it found the two similarly accurate for cup position and leg-length discrepancy.
The academic angle
In Scotland, Dr Joseph Mante, principal lecturer at Robert Gordon University, describes the UK picture as “developing, at best”. The UK currently regulates medical devices under the Medical Devices Regulations 2002, as amended, while the MHRA continues to reform the framework for software and AI as a medical device. New post-market surveillance requirements came into force in the UK in June 2025. Northern Ireland follows a different route, with the EU MDR and IVDR applying under the Windsor Framework. The EU AI Act can still matter to UK businesses. Its territorial scope includes providers outside the EU that place systems on the Union market, and providers or deployers outside the EU where the system’s output is used in the Union.
According to Mante, regulation must increasingly account for products that behave less like static devices and more like evolving systems. “Medical devices were largely static. Increasingly, assisted diagnostics, clinical decision support software, remote monitoring tools and connected devices operate as software ecosystems continuously interacting with patients, clinicians, datasets and cloud infrastructure,” he expands.
“AI accelerates this by enabling devices not merely to measure/record data, but to generate recommendations, predictions and prioritise decisions. Consequently, regulatory concerns increasingly focus on how decisions are produced and evolve, and how clinicians interact with them. The implication? Safety can no longer be evaluated solely at market entry. Regulators move towards continuous assurance throughout products’ life cycles, particularly where algorithms may change through updates or adaptations.”
That life-cycle emphasis is visible on both sides of the Channel. The EU AI Act requires post-market monitoring for high-risk systems. In contrast, the UK’s strengthened medical-device rules require manufacturers to gather and analyse post-market information and continue monitoring safety and performance.
For manufacturers, the question is how much existing quality infrastructure can absorb. “Many
13
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102
