Future quality audits of AI systems or their development in terms of AI quality management also make good sense and are already being considered in the EU’s AI Act. (Credit: wladimir1804/AdobeStock)

Thanks to artificial intelligence (AI), augmented reality (AR) has long shaped product development across a variety of areas, including the medtech industry. Use of these trends can significantly improve diagnostics and, therefore, treatment. This applies, for example, to surgery and to the adjustment of medication regimens to reflect the patient’s needs. To do this, medical practitioners use recommendations provided by AI, which in turn draws on a broad digital database.

Increasingly small and energy-efficient sensors are contributing to this and deliver a constant stream of information. They can be worn close to the body and monitor various body functions. Smartwatches already offer a host of these measurement options.

Tiny sensors, implanted into the body, take the process one step further. They use specific measuring techniques to monitor all body functions.

First products of this type are now used for conditions such as cardiac insufficiency by permanently monitoring the heart rate. Similarly, sensors are already in use that continuously measure blood sugar levels in diabetes sufferers. In both cases, the goal is timely initiation of appropriate measures to prevent the patients’ condition from worsening or potentially turning life-threatening. Other products monitor bone healing after fractures to provide individualized therapy.

Challenges: Data Protection and the MDR

Manufacturers developing these types of products need to consider a host of different aspects. First, they need to ensure compliance with data-protection regulations: Data must be stored and processed safely and protected from access by unauthorized parties. And manufacturers must also answer a question that constantly arises in connection with AI: are the AI’s decision-making capacity, accuracy, and correctness ensured? In this context, problems can arise if validation data sets lead to bias in AI. In preparing its Artificial Intelligence Act (AI Act), the EU is currently working actively on legislation. The requirements of the AI Act are likely to make the already extensive regulatory acts that govern medical devices even stricter.

Five years ago, the Medical Device Regulation (EU 2017/745) replaced the Medical Device Directives (93/42/EEC and 90/385/EEC) that had previously applied. Ever since, there has been a simmering debate over whether the provisions of the MDR are supportive of innovation or hostile toward it. Is the MDR’s improvement of patient safety based merely on the status quo, or does it also permit innovation? While this question cannot yet be answered conclusively, initial indications show that companies first tend to introduce their innovations in other markets such as the USA. The FDA has set up numerous programs to provide manufacturers as early as possible with information that may assist them with the successful development of medical devices. In addition, the FDA can always be contacted before an approval procedure to make preapproval arrangements and obtain technical assistance.

Currently, the EU has taken a different approach and has severely restricted all communication between Notified Bodies and manufacturers prior to a conformity assessment procedure. Notified Bodies are facing the risk of being classified as consultants and are thus very reluctant to take a proactive approach to this problem. The biggest challenges in this context are uncertainty and lack of transparency.

“Fail Early, Fail Cheap” — Early Tests Reduce Risks

What can manufacturers do today to improve their odds for successfully placing an innovative medical device on the EU market tomorrow? One option is to collaborate with accredited testing laboratories, which closely follow regulatory developments and can adjust to the latest requirements. These testing laboratories can test products from an early stage of development. Manufacturers that opt for such collaboration add certainty to their future conformity assessment procedure.

Well-known laboratory tests are already in place in fields such as electrical safety or biological compatibility. Less known but equally important is the topic of cybersecurity. A few accredited laboratories already offer tests in this area, scrutinizing medical devices for vulnerabilities. By pointing out areas in need of improvement, they help companies avoid having to remedy problems in the market under high (time) pressure and in the media spotlight. Future quality audits of AI systems or their development in terms of AI quality management also make good sense and are already being considered in the EU’s AI Act. TÜV SÜD is one of the few testing laboratories to offer this kind of service.

Failures that are identified at an early stage do not hold up the conformity assessment procedure. Of course, some challenges will remain, and early testing cannot eliminate all problems. However, while manufacturers will continue to face the challenging shortage of resources in the regulatory field, every test that is expertly performed and passed will lead to better predictability — and thus legal certainty.

This article was written by Martin Witte, Senior Director Strategic Business Development, TÜV SÜD Product Service GmbH, Munich, Germany. Contact: This email address is being protected from spambots. You need JavaScript enabled to view it.. For more information, visit here .