EU Commission

New EU Study on the Deployment of AI in Health

The European Commission has released its Study on the Deployment of AI in Healthcare, a comprehensive look at how artificial intelligence is being introduced across Europe’s hospitals and clinics. The report shows that AI is already improving efficiency, diagnostics, and equity of access. At the same time, deployment remains uneven, slowed down by a mix of technical, legal, organisational, and cultural challenges.

Early Signs of Impact
The study highlights areas where AI is already delivering results:

  • Efficiency. Digital scribes and natural language processing (NLP) are cutting documentation time and giving clinicians more patient-facing minutes. Predictive analytics are helping hospitals reduce waiting times and manage patient flow. NHS England, for example, is expanding AI to address missed appointments and scheduling delays.
  • Diagnostics. Radiology, pathology, oncology, cardiology, and stroke care are at the forefront. AI tools have shortened turnaround times for critical CT scans, improved cancer detection, and enabled earlier interventions when every minute counts.
  • Equity. AI is helping overcome shortages of specialists by enabling remote diagnostics for conditions like tuberculosis and cancer. Telemedicine platforms and AI-supported monitoring bring care closer to underserved populations, while virtual training systems expand workforce capacity in rural regions.

Four Barriers to Deployment

If the benefits are clear, why isn’t AI scaling faster? The study groups the obstacles into four categories:

  • Technical. Fragmented, unstructured data and a lack of interoperability remain major hurdles. Many hospitals still lack the digital infrastructure to support large-scale AI deployment.
  • Legal and regulatory. Complex approval processes, questions of liability, and uneven implementation of EU rules add layers of uncertainty. National variations further complicate cross-border scaling.
  • Organisational. Hospitals face unclear financing models, limited capacity to integrate AI into existing workflows, and uneven levels of digital maturity. Without stable governance and budgets, projects often remain stuck in “pilot mode.”
  • Social and cultural. Trust in AI is still fragile. Clinicians worry about reliability and workload, patients about transparency and data use. Limited digital literacy across both groups slows down adoption.

Together, these barriers show that the problem is not algorithmic performance but system performance.

Building Smarter Systems

The Commission’s study also points to accelerators that could help Europe move from pilots to practice: common standards for interoperability, centres of excellence, clearer financing models, assurance labs for local testing, and a EU-wide catalogue of validated AI tools. The message is clear: Europe’s challenge is not to invent more AI, but to build the systems that allow it to thrive. If policymakers, hospitals, and innovators succeed, the prize will be profound – faster and more accurate diagnoses, professionals freed from paperwork, and healthcare that is fairer, more efficient, and more accessible for all.

Source: Study on the deployment of AI in healthcare

Yiannos Tolias, Legal Lead on AI and AI Liability in Healthcare & EHDS Team at DG SANTE, led the study from the commission. He will be speaking at the Data2Value Executive Dialogue on October 14, 2025, in Berlin.


Data2Value | Executive Dialogue
Envision a World Where Every Healthcare Decision is Informed by Data.