On 29-30 April 2026, several expert-level meetings, supported by the Cyprus Presidency of the European Union at the closing conference of the Xt-EHR joint action, took place on important AI-related topics. The chief subject of discussion was about building a robust and secure AI-enabled connected Health Data Space.

Source: Cyprus Presidency

The EHDS will absolutely need to demonstrate its success through bringing concrete health benefits both to citizens and to the European Union as a whole.

Hence, the events brought together stakeholders from health and digital health sectors all across the European Union. The discussants highlighted the dual visions of the European Health Union and the EHDS, while they promoted cooperation and the integration of health systems. Speakers focused on strategies to address challenges in the health sector, the European Health Data Space (EHDS), at the same time as the rapidly evolving application of AI in healthcare. AI-specific topics included Explainable Artificial AI, Trustworthy AI, and AI-driven decision support. These topics were raised alongside the need to enhance digital system interoperability.

Held in the context of the Xt-EHR joint action closing conference, representatives of the TEHDAS2 joint action were also present. Clearly, European joint actions benefit substantially from their mutual collaboration. EHTEL’s digital health facilitator, Luc Nicolas attended the event remotely, and offered EHTEL the following feedback.

What was new and what were some key positive messages?

Overall, many of the speakers emphasised the need to connect both the primary and secondary uses of health data. Indeed, most speakers claimed that AI is a key enabler for EHDS success.

Several important messages were offered:

• A number of large-scale European Union investments have already been made. Several programmes and initiatives are already active. Angelo Marino of the Union’s HADEA agency announced that: more than €110 million in funding has been made available, and 77 projects have built the groundwork that helps construct the EHDS. Multiple programmes are now aligned in their dedication to both the EHDS and AI (they include EU4Health, Horizon Europe, Digital Europe, and the Connecting Europe Facility).
• AI already delivers measurable value in the fields of health and care. Useful examples cited by Fulvia Raffaelli of the European Commission include faster diagnosis (e.g., of pulmonary embolisms), reductions in administrative burden (of ~17%), and solid predictive hospital management.
• EHDS implementation is based on a learning cycle. This “EHDS learning cycle” involves the integration of policy, technology, governance and stakeholders.
• Europe is now politically and strategically aligned on the EHDS and AI. There have already been many positive achievements based around legal frameworks, funding, and the early set-up of infrastructure.

Indeed, the EHDS is no longer a policy initiative – it is now a system transformation challenge. Essentially: “We are no longer designing policy, but executing it.”

EHDS success will be judged not by regulation, but by measurable improvements in patient care and health system performance.

So, what is needed in terms of primary and secondary uses of data?

Historically, the two uses of health data (primary and secondary) were kept separate. This means that today systems are too fragmented. This situation must change.

Integration has to be the central goal. Integration requires aligned governance, not just technology.

Essentially, a single governance model – which is supported by common infrastructures – could focus on both primary and secondary data uses.

Secondary data, of course, however, does differ from the primary use of data in at least one significant way – in the number of its datasets and categories: “Secondary use requires more data types, more standardisation, and more governance efforts.”

Yet, effectively, in total, it’s all about “data for the citizen”.

Bringing primary and secondary uses of data together is therefore “a system of bridges”, it’s “not a single bridge”. Just as primary use data feeds forward into secondary use data, the secondary use of data feeds back into primary use of data. Both types of data can operate well together.

The ultimate aim is to offer people better healthcare.

What is still missing and what challenges remain?

Four important initiatives are not yet in place. They are:

• Real-world implementation at scale
• Trust and maturity of governance mechanisms
• Workforce readiness
• Integration across systems

Some challenges definitely remain, and there is a need to tackle them. Here are some examples:

• Public health data: There is need to overcome the lack of clarity regarding public health data. As one speaker said, “Data can be used for other purposes [than simply public health]. We need a shared common understanding … to avoid the misuse of data.”
• Legal and ethical definitions: Across European countries, some legal and ethical definitions are still unclear and inconsistent.
• Technical alignment: While standardisation and interoperability are key enablers, technical alignment is often still debated.
• The once-only principle: The secondary use of data must not increase people’s workload; its use should be automated. As a result, the “once-only principle”  should be emphasised. As another speaker said: “Doctors … have to manually register data … double or triple register data … that’s crazy. We need to move toward automatic data flows.”
• AI is essential: Last but not least, AI can “help structurise data” as well as encourage the use of unstructured data.

Some last reflections on AI

Hence, AI is essential for scaling up secondary use of data, especially when dealing with messy, real-world data.

While there is great room for optimism, for example, about AI (and the transformation, efficiency, and innovation that it can bring), the approach to it needs to be tempered with caution (about the potential risks related to bias, inequalities, and the need for a genuine materialisation of trust).

This identification of what is needed in terms of the integration of primary and secondary uses of data, AI, and important remaining challenges show that there are many concrete areas in which implementers voices need to be heard, and on which implementers can work.