Artificial intelligence in medical devices: authorities issue guidance
Federal Network Agency, Federal Commissioner for Data Protection and Freedom of Information and Hesse issue a joint “Roadmap” – which needs to be adjusted soon.
(Image: Whyframe/Shutterstock.com)
For medical devices such as pacemakers, insulin pumps, electric wheelchairs, but also health software like Digital Health Applications, learning models play an increasingly important role. However, especially in this area, Artificial Intelligence is considered particularly sensitive, as errors can be life-threatening. Several official German bodies have now jointly compiled an orientation guide for manufacturers and providers on what they need to pay attention to. The roadmap is based on the findings from the so-called AI real-world laboratory process: In a controlled test environment, two start-ups tested possibilities and limits within the regulatory framework.
All companies in the German medical device industry should be able to benefit from the orientation guide. “Anyone developing AI in medicine needs clear guardrails, not regulatory uncertainty,” says Louisa Specht-Riemenschneider. The Federal Commissioner for Data Protection and Freedom of Information, who will be stepping down from her post due to illness, sees this as an important step: “This enables good innovation: with more legal clarity and in the interest of patients.” This example shows “how important the dialogue between companies and authorities is to shape practice-oriented regulation,” says Klaus Müller, President of the Federal Network Agency, whose authority is also to act as a supervisory authority for the AI regulation, while medical devices are otherwise subject to the supervision of the Federal Institute for Drugs and Medical Devices and state authorities.
Hesse's Minister of Digital Affairs, Kristina Sinemus, points out that medical devices are particularly dependent on high-quality data – and is also setting up a suitable institution in her own federal state with the Trust Agency for Medical Data (MDTS).
Videos by heise
Politically, the utilization of initial results from real-world laboratories is important, as they are considered a middle ground between prohibition and preventive full regulation. The guide, now named “Roadmap,” is primarily a highly condensed representation of existing obligations and, depending on the case constellation, possibly relevant regulations to be observed. Due to the interactions between various regulations, these are by no means easy for providers to understand and are also still in flux. The extent to which the topic is still in motion is shown by the timing: while the assistance was finalized at the end of April, work was still being done in Brussels on the Omnibus Act for the AI regulation – and an agreement was reached last night between the European Parliament, the Council of Member States, and the Commission. A revision of the presented “Roadmap” document is therefore very likely to follow soon, once the changes to the legal text of the AI regulation have also been formally adopted.
(mho)
