Logo of the Physikalisch-Technische Bundesanstalt
Into the Future with Metrology - The Challenges of Medical Technology

Digitalization in the Health Care System

Matters close to the heart: artificial intelligence

ECG database

Cardiovascular diseases are among the most frequent causes of death worldwide. Consequently, examinations using electrocardiograms (ECGs) are a crucial aspect of healthcare. With the introduction of telemedicine, e.g. for monitoring long-term ECGs, the importance of such examinations will continue to grow. Automatic ECG interpretation algorithms based on Deep Learning can significantly reduce the workload of medical personnel. However, most algorithms have only been trained on non-public data sets. These data sets are usually not large enough to be reliable for this purpose. Furthermore, the evaluation methodology has not been standardized, so that results cannot be sufficiently compared. PTB has solved this problem by developing a large, publicly accessible database. It contains 21,837 so-called 12-lead ECGs of 18,885 patients and is the largest public clinical data set of its kind: about 40 times larger than the PTB Diagnostic Database used so far. The database provides machine-readable diagnostic findings and over 70 different annotations on the ECGs from up to two cardiologists. Furthermore, the data set contains many co-morbidities in addition to data of healthy patients. Healthy patients’ data are often underrepresented in clinical data sets. Different signal qualities are also included. The database is therefore ideally suited for training and evaluating machine-learning algorithms on a real-world data set.

 

Division involved

Opens internal link in current window8 Medical Physics and Metrological Information Technology