That was perhaps the real question we tried to answer earlier this week in a live webinar with Royth von Hahn (TÜV SÜD) and myself under the wing of Andrea Pietsch (TÜV SÜD) as moderator. Many topics passed the floor, including testing for device safety and compliance, the MDR transition process, and the new annex XVI products, but above all the impact on innovation and smaller companies being able to continue in a landscape of growing regulatory complexity and overall compliance burden.
Wearables and AI-enabled devices were discussed, but also neuro-stimulation, connected devices, 3D-printed products, and personalized medicine. But on top of that also the wider perspective in for example brain surgery. More thorough scans and AI-powered diagnosis, robot-assisted surgery, and precision medicines together are probing into this largely undiscovered area. And good to observe that these new technologies once well validated, appear to have an acceptable throughput time in the notified body conformity assessment. So, in that sense innovation is fostered in the EU.
We hear about perceived and real barriers for innovative devices to enter the EU market, but to me, that sounds like a challenge. Once you are aware at the start of what hurdles the obstacles race will bring, with good preparations you might very well reach the finish line. The process from idea to practical clinical application is indeed long and expensive. Traditionally, early research is performed in academic institutions, while device development, testing, and production occur in the corporate environment. Processes are costly and frequently take years to accomplish. Historically speaking, Europe took an easy path to clinical validations, and hence the obvious first go-to market is shifting, as this aspect has been strengthened considerably in the new legislation. Legislators around the world take a keen interest in healthcare market sectors, and new entrants may have to jump over several barriers before they can enter any healthcare market. New market entrants face additional financial and administrative challenges when it comes to licenses, permissions, compliance, etc. The best way to approach all these hurdles is to have a detailed regulatory and clinical strategy, choosing the first go-to countries based on phased validation methods. So, a stepwise approach in which the streamlining of clinical testing, including a more standardized approach to the health technology assessment process, might speed up the introduction and spread of such innovative devices.
As Royth alluded to, the innovations share these challenges with the so-called orphan devices resp. niche applications. The only difference is that innovations usually are designed to scale in a later phase of their product life cycle while for orphan devices these challenges are persistent.
We touched upon the problem of PFAS, abundantly used in devices, but their long-term safety forces a material innovation. Many are on to this quest. The biggest concern I have as a chemical engineer is that we would replace products, where we know their risk profile, and know their challenges and limitations, by products where many more aspects are unknown or at least less fully understood. So, the core question will always remain: is the alternative safer? And that means serious in-depth testing of any new and innovative material before its use in medical devices.
A key challenge faced by medical technologies with integrated AI capabilities is their ability to adapt their prediction to reflect accumulated data. Traditionally, the assessment of the safety of medical devices has been based on predetermined and clearly defined risk assessment principles and practices. However, by design, many algorithms and data models used in certain AI-enabled medical technologies are not “locked” but instead continuously learn and adapt their functionality in real time to optimize overall performance. As a result, these technologies may well present one risk profile during the initial product development process and a different risk profile after the device has been deployed for use with patients. In this context, organizations developing medical technologies with integrated AI capabilities should strongly consider taking a more expansive approach in assessing the safety of their products. Such an approach would address every aspect of the product planning and development process and extend beyond the initial product release date to include rigorous post-market surveillance activities. The safe choice sometimes adhered to is a stepwise validation, where the AI continues to learn in a sandbox environment. But that of course limits the capabilities and speed of improvement of the devices and somewhat reduces the full power of innovation.
In all of this, the collection of sufficient clinical data in both the pre-market as well as the post-market phase are perhaps the greatest hurdles in the sense that they absorb a lot of time, energy and money, and getting it right is difficult but essential. Pre- and post-market clinical studies are key factors in demonstrating device safety and compliance. I’d say in terms of validation, these are the proof of the pudding. Is the product delivered and is it as safe as we have considered it to be throughout the development phase? Pre-market clinical studies allow for early identification of risks, efficacy assessment, and regulatory compliance. But its value is limited by the parameters set in the (one could say) artificial setting, with inclusion and especially exclusion criteria. Therefore, post-market clinical studies are essential in monitoring the long-term safety and performance of medical devices, more in a setting of how they are used in the real clinical setting; they are also extremely helpful in the identification of emerging issues and finding unforeseen risks.
The approach Qserve takes in supporting the generation of clinical data is an integral one. Integral in two ways:
- First and foremost, the clinical data collection, the tools, its timing, and the subsequent order in which various types of data collection are included in the life cycle management of products in our minds should be fully integrated into the regulatory strategy pre-market and the regulatory reality post-market. Only if the clinical data are fully aligned with the further validation and evidence in the dossier is it truly useful and hopefully sufficient clinical evidence.
- Secondly, we believe in working together. Just outsourcing a clinical study to a CRO might seem easy, although perhaps expensive. However, in many cases, the data derived from a clinical study does not match the regulatory reality of the further dossier. Hence, we strongly believe in a partnership between the company and the CRO.
Especially also for startups, and scale-ups, struggling to allocate vast amounts of currency, such an integral approach has many advantages in supporting a potentially faster approach towards market entry.
Coenraad Davidsdochter: ‘’Following the discussion, it became apparent that the regulatory strategy for compliance will and should play a much more prominent role because of the enlarged differences between the EU and the US. Predictability has been the keyword mentioned for manufacturers to base their decisions on regarding such a strategy. Especially regarding AI and Machine learning, the EU is introducing another layer of unpredictability with the new horizontal AI ACT that is expected to be published in early 2024 and that will hold additional requirements for medical device manufacturers and potentially numerous additional horizontal and sometimes conflicting horizontal standards to the AI ACT to be met.’’
Davidsdochter continues: ‘’I enjoyed the lively discussion and was glad to find out that efforts to clarify and help medical device manufacturers understand the regulations are not only of the highest importance for consultancy’s like Qserve, but also under the attention of individual Notifies Bodies like TÜV SÜD.’’
For more information about Qserve’s services for SME’s, feel free to check out our website. For more information about our clinical services, please visit Qserve CRO.