Metrology of automated data analysis for cardiac arrhythmia management

Short Name: MedalCare, Project Number: 18HLT07
Image of heart beats on a ECG
Heart monitor vector

Validating software for automatic diagnoses of cardiovascular diseases

Cardiovascular disease (CVD) is responsible for 3.9 million deaths a year in Europe. Currently, Electrocardiography (ECG) is used for a non-invasive and cost-effective way for initial clinical examinations and subsequent patient monitoring. Automated detection systems and computer-based machine learning techniques are becoming available for diagnosing and monitoring CVD such as ischemia and arrhythmia. To build trust in automated CVD diagnostics, and help reduce healthcare costs, a standardised procedure needs development to validate complex underlying algorithms and machine learning techniques.


This project developed a synthetic reference ECG measurement dataset, including healthy variations and selected CVD pathologies, to performance test CVD diagnostic devices. The project has, for the first time, provided traceability for CVD data analysis techniques. Such standardised testing will help manufacturers develop new ECG devices with improved CVD diagnosis reliability, thus helping promote uptake of the technology, both in clinical use and for monitoring equipment for use in the home.


Project website
Robustness of convolutional neural networks to physiological electrocardiogram noise

Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences

Deep Learning for ECG Analysis: Benchmarks and Insights from PTB-XL

IEEE Journal of Biomedical and Health Informatics

Other Participants
Arrhythmia Alliance (United Kingdom)
Fraunhofer-Gesellschaft zur Foerderung der angewandten Forschung e.V. (Germany)
Karlsruher Institut fuer Technologie (Germany)
King's College London (United Kingdom)
Medizinische Universität Graz (Austria)
Technische Universität Berlin (Germany)