Unlocking hospital data for cardiology: DataTools4Heart platform enters clinical testing phase

Sophia Antipolis, France – September 2026: Cardiovascular disease (CVD) causes one in three deaths across the European Union, making it the leading cause of death in Europe. The disease costs the EU an estimated €282 billion annually. 

European hospitals generate large volumes of cardiology data during routine care, but much of it is not reusable for research. The data are held in different formats, recorded in different languages, and stored in systems that were not designed for secondary use. Many records are stored in unstructured clinical notes, and the lack of interoperability between systems further limits the potential for data-driven research and innovation. 

DataTools4Heart (DT4H) is a Horizon Europe Research and Innovation Action that addresses these barriers by developing a federated, privacy-preserving cardiology data toolbox. Rather than moving patient data between institutions, the DT4H approach keeps data at the hospital and moves the analysis to it. 

The project has now reached a pivotal milestone: the platform is being tested with real clinical data at seven hospitals across Europe. This marks a significant step towards making cardiology data more accessible for research and care, while ensuring patient privacy and data security remain paramount. 

The platform integrates: 

• Standardised data ingestion and harmonisation tools based on a common data model 

• Multilingual natural language processing to extract information from unstructured clinical text 

• Federated machine learning that trains models across sites without centralising patient data 

• Differentially private synthetic data generation (CardioSynth) 

• Language models tailored to the cardiology domain 

• Virtual assistants to help clinicians and researchers navigate multi-source cardiology data 

All tools are being developed with privacy-by-design, in compliance with European regulations and data standards, and are validated at seven clinical sites in seven countries (Netherlands, Sweden, United Kingdom, Italy, Spain, Czech Republic, and Romania). 

The testing phase is crucial for measuring the platform’s performance with the irregularities and gaps found in real-world hospital records. The data ingestion suite has been enhanced to identify inconsistencies and coding errors early, and the onboarding process for hospitals has been simplified. 

“Having the federated learning backend running against real hospital data is the step that turns an architecture into a usable platform. What we learn from these first tests, particularly on data quality and missingness, will determine how the toolbox performs at the remaining sites,” says Karim Lekadir, ICREA Research Professor at the University of Barcelona, Director of the BCN-AIM lab and coordinator of DataTools4Heart. 

The consortium is also evaluating several imputation strategies to improve dataset completeness without compromising quality. The European Society of Cardiology represents the perspective of the cardiology community within the consortium. ESC volunteers and its network of national cardiac societies contribute to clinical requirements, review the usability of the tools against everyday clinical practice, and support the dissemination of project results to clinicians and researchers across Europe.  

Link to the ESC press release