A fully automatic and real-time arrhythmia classifier

Institute for Cross-Disciplinary Physics and Complex Systems (IFISC)

The electrocardiogram is the most powerful and key tool to diagnose cardiovascular diseases. IFISC researchers developed a fully automatic and real-time arrhythmia classifier based on a single ECG lead. The classification performance of the proposed model outperforms previous single-lead arrhythmia classifiers, achieving 85% success rate in patients with arrhythmia and 98% success in healthy subjects. These results are comparable to those obtained with state-of-the-art algorithms using multiple leads. The proposed method uses an Echo State Network (ESN) to classify ECG signals following the Association for the Advancement of Medical Instrumentation (AAMI) recommendations with an inter-patient scheme. In addition, the approach allows toi transfer the knowledge from one database to another without additional training.

Example of an electrocardiogram obtention method: the Holter monitor.

Part of the study was carried out in collaboration with the company Nuubo, a company with more than 10 years of experience in the development of biomedical solutions in the field of wireless monitoring using new materials. Nuubo has been pioneering in developing wearable devices to analyze heart activity. IFISC and Nuubo signed a collaboration agreement in 2015 to join forces and apply the IFISC group's knowledge of neuro-inspired algorithms to ECG classification to improve the diagnostic performance of the company's software. As a result, the system developed by IFISC and Nuubo was patented both in Spain and US in 2017.

In September 2018 Nuubo received approval from the U.S. Food & Drug Agency to market its product in the US. The company anticipated its entry into the American market in the fourth quarter of the year 2018.


Image credits:

Electrocardiogram pictures at article frontpage is in the public domain and was downloaded from Maxpixel.

In-text image of Holter monitor is in the public domain and was downloaded from Wikimedia Commons.