Blog - General

Legal and ethical frameworks for AI are often created at a distance from the environments where AI systems are actually developed. While these regulations aim to ensure fairness, transparency, and safety, developers frequently experience a gap between regulatory expectations and the practical realities of building AI models.

As part of WP1, discussions with AI development teams explore how legal and ethical requirements are experienced during the development process, particularly in healthcare and data-driven research.

One key theme is that developers generally understand the importance of regulation, especially regarding patient protection, fairness, and accountability, and requirements around explainability, documentation, and bias mitigation are to a large degree perceived as well aligned with scientific quality expectations.

Data-related requirements are among the biggest concerns. Ensuring data quality, completeness, representativeness, and security is critical for trustworthy AI, but real-world healthcare data is often incomplete, fragmented, or potentially biased. Developers therefore rely on mitigation strategies such as transparency about limitations, human oversight, and continuous monitoring.

Legal requirements also increasingly shape model design itself. Expectations regarding automatic logging, transparency, and human oversight are generally viewed as valuable safeguards, but they can also introduce technical complexity and slow development processes.

Questions about responsibility, acceptable risk, and fairness remain difficult to answer in practice. The discussions further highlight the importance of involving clinicians, patients, and other stakeholders in AI development. Including those who will ultimately use or be affected by AI systems can improve both ethical robustness and practical usability.

Overall, the conversations within WP1 demonstrate that ethical and legal frameworks are becoming an integral part of AI development. The challenge moving forward is to ensure that these frameworks not only protect individuals and society, but also remain realistic, understandable, and adaptable to the realities of innovation.

In an amazing setting, Villa San Remigio in Verbania with beautiful views over Lake Maggiore  in northern Italy, we recently spent 4 days attending a course learning more about computational tools for prediction in medicine.

The course covered the whole journey from idea to deployment:

- how to find out what needs stakeholders (patients and clinicians) have,

- how to design a tool that meets these needs,

- how to address ethical and legal regulations when handling sensitive date

- how to design tools that are fair

- how to handle IP rights

- how to harmonize data

- approaches for sharing sensitive data, making it FAIR

- different computational approaches for prediction using health data

We also had several inspirational talks:

Robot automation in Rheumatoid Arthritis diagnosis – the Arthur Ultrasound AI Robot by Dr Rajeeth Savarimuthu University of Southern Denmark

Addressing Bias, Generalizability, Explainability, Interpretability by Professor Peter Elkin  University of Buffalo

From Graph to LLMs; from Bioinformatics to equivariant properties by Prof. Pietro Lio Cambridge University

The students developed ideas for predictive tools and assessed ethical compliance as well as made plans for carrying out such a project taking all aspects they had learned into account. We were very impressed by the design of their project and how much of what was taught that they had managed to incorporate in their project descriptions.

The students also presented their research from their PhD or post doc studies. The poster session was full of discussions with new ideas and connections being formed.

The beautiful, quiet environment was perfect for learning and interactions between participants. We also got plenty of exercise every morning walking up a hill from the hotels to the Villa San Remigio where we held the summer school.

Thank you so much to all the students and teachers for making this a memorable occasion from which we all came home with new knowledge.

If you wish that you had been part of the course, please let us know – we are exploring possibility of offering it again. Send an e-mail to: wisdom.horizon.2023@gmail.com

On 19 March 2025, the European Multiple Sclerosis Platform (EMSP) hosted a dynamic consultation meeting with the Multiple Sclerosis Community Advisory Board (MSCAB) and researchers from the WISDOM project. The meeting was part of the project’s Gate 1 co-design phase, where nine innovative digital tool ideas were reviewed and discussed.

The goal? To ensure that the development of digital health tools for MS is guided by real-world patient expertise and insights.

As one CAB member shared, “I felt very included and appreciated the diversity of perspectives.” Another added, “The meeting was engaging and respectful. It made me feel that our opinions genuinely matter.”

A Lively Dialogue with MS Advocates

MSCAB members — people living with MS from across Europe — shared their diverse perspectives on tool feasibility, relevance, and usability. Discussions were lively and inclusive, reflecting both the hopes and the concerns of patients who are eager to see new tools that truly meet their needs.

One participant noted, “It was truly a dialogue – not just a presentation.”

Top-Ranked Tools: Personalised, Practical, Patient-Focused

The CAB’s voting exercise highlighted a clear preference for tools that promote personalised care and support shared decision-making. Ideas like a personalised intervention strategy, disease progression trajectory, and personalised intervention monitoring received the most enthusiastic endorsements.

At the same time, participants voiced thoughtful caution around predictive tools that might cause anxiety if not accompanied by clear clinical context. Data privacy, ethics, and risk communication were recurring themes in these discussions.

As Professor Uffe Kock Wiil from the WISDOM project team shared:

 “I have recently interacted for the first time with the EMSP Community Advisory Board (CAB) in relation to development of AI tools for MS decision-making in the context of the WISDOM Horizon Europe project. I did not know what to expect, but I was very pleasantly surprised. The feedback from the CAB members was very useful for our future tool development process, and I already look forward to the next CAB meetings.”

One CAB member summed up the atmosphere perfectly: “The openness to feedback and the researchers’ attitude was very reassuring.”

Key Takeaways and Next Steps
The CAB’s recommendations will directly inform the WISDOM Assessment Panel’s decision on which tools to advance to concept development. These recommendations include prioritising patient-centred design, embedding ethical safeguards, and ensuring that digital tools integrate seamlessly into real-world clinical care.

As another participant put it, “My expectations were exceeded – I learned a lot and also felt that my input will impact the project.”

Please Note:
While the detailed meeting report remains confidential to protect participants’ privacy, we are committed to sharing project updates and key learnings as the WISDOM project progresses.

The WISDOM project team is also committed to continuing this co-design journey, with a follow-up consultation planned in 2026 to ensure that patient voices remain at the heart of innovation.

More Information
For updates on the WISDOM project, visit https://wisdomhorizon.eu/ .

More information about MSCAB: https://emsp.org/projects/ms-cab/ .