Europe wants to accelerate AI adoption in healthcare

Europe is preparing a new push to accelerate the use of artificial intelligence in healthcare, moving the debate beyond regulation and towards large-scale implementation. Healthcare has been named as one of five high-value sectors in which the European Commission sees particularly strong potential for industrial AI, alongside transport, agri-food, advanced manufacturing, and defence and space. New European initiatives targeting these sectors are expected to be announced in November.

The shift was set out by European Commission President Ursula von der Leyen during her 2026 State of the Union address on 16 September. Her central argument was that Europe's next AI challenge is not simply developing more advanced models, but turning the technology into measurable value in the real economy and public services. In healthcare, that means moving AI closer to clinical workflows, while retaining human responsibility for diagnosis and treatment.

Breast cancer screening was used as a concrete example. AI-supported mammography can help detect cancers during screening, but the Commission's position is that such systems should strengthen the capabilities of healthcare professionals rather than replace them. That distinction is likely to become increasingly important as Europe attempts to combine faster AI adoption with the safeguards established through its regulatory framework.

From regulating AI to deploying it

Europe has spent much of the past several years building the rules and infrastructure surrounding artificial intelligence. The AI Act provides a common regulatory framework, while the European Health Data Space (EHDS) is intended to make health data more accessible and interoperable for healthcare, research and innovation. At the same time, Europe is investing in computing capacity, AI factories, health-data infrastructures and testing environments designed to help organisations develop and validate new applications.

The next challenge is connecting those elements to clinical practice. The Commission itself acknowledges that despite substantial progress in developing AI and machine-learning-enabled medical technologies, adoption in routine healthcare remains relatively slow. Technical performance is only one part of the problem: organisations also face regulatory uncertainty, fragmented data, integration with clinical workflows, financing questions, workforce requirements and the need to demonstrate value in real-world healthcare environments.

Several European programmes are already being built around those barriers. The Commission's Apply AI Strategy focuses on accelerating practical AI use in strategic sectors, including healthcare and pharmaceuticals, while AICare@EU addresses obstacles to deploying AI in clinical practice. Europe is also establishing a network of AI-powered advanced screening centres intended to validate technologies in real healthcare environments and generate evidence that can support adoption across national health systems.

The ambition is substantial. The screening network is initially focused on cancer and cardiovascular disease and is intended eventually to cover all EU Member States. Rather than simply demonstrating that algorithms can perform well in research settings, participating centres are expected to assess clinical performance, organisational impact and integration into existing care pathways.

Health data becomes strategic infrastructure

Europe's healthcare AI ambitions are closely tied to its health-data strategy. The Commission argues that European healthcare systems hold large quantities of high-quality data that could support the development of specialised AI models, provided those data can be accessed and used securely. The EHDS is therefore becoming more than an interoperability project: it is also part of the infrastructure Europe expects to use for AI development, validation and deployment.

Several large-scale initiatives illustrate the direction. Cancer Image Europe is intended to provide access to 60 million cancer images by the end of 2026, creating a cross-border resource for developing and testing imaging algorithms. Twenty-six countries are participating in the 1+ Million Genomes initiative, while a separate European infrastructure is being developed for intensive-care data that can support predictive models, clinical decision support and AI-based risk prevention.

Those programmes matter because the performance of medical AI depends heavily on the quality and diversity of the information used to develop and evaluate it. A model trained within one hospital or on a narrow patient population may not perform equally well when introduced elsewhere. Cross-border health-data infrastructure could make it possible to test technologies against more representative populations before they are deployed at scale.

The challenge will be turning access to data into demonstrable improvements in healthcare. Large datasets and computing infrastructure do not solve implementation problems on their own. Hospitals still need technical integration, clinical governance, sustainable financing and professionals who understand how AI should be used and when its recommendations should be questioned.

The next test is implementation

The Commission's decision to place healthcare among its five priority AI sectors therefore comes at an important point. Europe has spent years establishing regulatory guardrails and building digital infrastructure; it is now putting greater emphasis on translating those foundations into practical adoption. The initiatives expected in November should provide a clearer indication of how far the EU intends to go in supporting that transition and what role healthcare organisations, Member States and industry will be expected to play.

For health systems, the relevant question is no longer whether AI will become more capable. It is whether healthcare organisations can introduce those capabilities safely into real clinical pathways and demonstrate that they improve quality, accessibility, productivity or outcomes. That requires considerably more than buying software: workflows may need to change, professionals need appropriate skills, data must be available and reliable, and responsibility for AI-supported decisions must remain clear.

Europe's emerging approach increasingly connects those pieces. The AI Act establishes safeguards, the EHDS provides a framework for health data, European computing and testing infrastructure supports development and validation, and deployment programmes are intended to bring technology into healthcare practice. Whether those elements can work together quickly enough to produce large-scale clinical adoption will now become the more important test.

The Commission's message is therefore significant beyond the political announcement itself. After years in which European healthcare AI was frequently discussed through the lens of rules, risks and future potential, the emphasis is shifting towards implementation. The next phase will have to demonstrate not only that Europe can regulate trustworthy healthcare AI, but that it can actually put it to work.