ERC report shows frontier AI research can help deliver EU rules on trustworthy AI in health

238 ERC projects using AI in health

A new Feedback to Policy report analyses 238 ERC projects using AI in health, funded under FP7, Horizon 2020 and Horizon Europe, with a total budget of EUR 450 million. The projects have used AI for disease prevention and early detection, diagnosis, treatment optimisation and long-term disease management, and develop AI models, clinical decision-support systems and platforms, including machine learning and deep learning.

The study shows how AI-based models, clinical decision-support systems and platforms – including machine learning and deep learning – are being developed to enable earlier disease detection and more personalised risk prediction, diagnosis, prognosis and treatment. It also highlights how AI supports the integration of multi-omics, phenotypic and health data, and contributes across the medicine’s lifecycle, from drug discovery to clinical trials.

Links to the EU AI Act and European Health Data Space

The report shows how ERC projects can support the implementation of the EU AI Act, which classifies most AI-based software intended for medical purposes as ‘high-risk’, as well as the European Health Data Space and the EU’s Apply AI Strategy. ERC-funded researchers highlight the need for rigorous validation, robust risk management, high-quality data, transparency and meaningful human oversight, alongside secure infrastructures and clear data governance.

Gerd Gigerenzer, former Vice-President of the ERC Scientific Council, said:

Our analysis shows that real impact depends not only on better algorithms, but also on how they are designed, validated and governed. Smart technologies require smart institutions: without high quality data, transparent models, meaningful human oversight and clear accountability, the full potential of AI in health will not be realised.
 

Case studies, enablers and next steps

A more in-depth look at 59 projects and 20 case studies illustrates applications in disease detection and monitoring, drug discovery, risk forecasting, imaging, medical robots and personalised medicine – and points to long-term funding, AI-for-science hubs and regulatory sandboxes as key enablers.

The report demonstrates how frontier research can help ensure that AI in health is not only innovative and competitive, but also trustworthy, human-centric and firmly grounded