How can innovative AI applications in healthcare be developed and brought into clinical practice while meeting increasingly complex regulatory requirements? This was one of the central questions behind avisé labs’ participation in a nationwide pilot project for AI regulatory sandboxes.
In a new podcast episode, our CEO Marius Khan discusses his experience after five months in the pilot project, which was conducted by the German Federal Network Agency, the Hessian Ministry for Digitalization and Innovation, and the Federal Commissioner for Data Protection and Freedom of Information. Representatives from TÜV and the German Federal Ministry of Health were also involved.
For avisé labs, one key challenge was a familiar chicken-and-egg problem: without certification, access to clinical environments can be difficult, while without access to real clinical data, obtaining certification can be challenging.
The regulatory sandbox could not fully resolve this issue. At the same time, the discussions highlighted open questions and areas of tension between different regulatory frameworks, particularly the EU AI Act and the Medical Device Regulation (MDR).
Nevertheless, the participation provided valuable insights for avisé labs and helped identify practical approaches for both companies and regulators.
Three lessons stood out in particular:
- Start early: Regulatory requirements should be considered from the product idea and architecture stage, rather than only towards the end of development.
- Provide practical guidance: Checklists, standards and concrete recommendations are more useful to companies than extensive summaries of regulatory requirements.
- Move at startup speed: Regulatory feedback needs to be provided within a timeframe that allows innovative companies to keep moving forward.
Germany was among the early countries to gain practical experience with AI regulatory sandboxes. By August 2027, all EU Member States will be required to have at least one AI regulatory sandbox. At the same time, the German Federal Ministry of Health is developing further initiatives specifically focused on AI regulatory sandboxes for healthcare.
In the podcast, Marius Khan discusses what these experiences mean for the future of AI regulation in healthcare and what regulatory sandboxes need to provide to become genuinely useful for innovators.
Listen to the full episode on Spotify:
Listen to the podcast on Spotify
Many thanks to Marcel Holick for the invitation and the insightful conversation, and to everyone involved in the AI regulatory sandbox pilot for the constructive exchange.