A non-invasive method for early sudden cardiac death detection.

Abstract

Sudden cardiac death (SCD) is considered one of the main causes of mortality worldwide. Frequently, people with a history of cardiopathies exhibit the symptoms. However, one hour prior to the fatal occurrence, the symptoms can also appear in healthy individuals. Understanding the origin of this heart disease continues to be a challenge for the scientific community. According to the state of the art, new methods to stratify and predict SCD have been proposed. Different algorithms have been implemented, and risk stratification indices have been used, such as heart rate variability (HRV) and T-wave alternans (TWA). TWA is the term used to describe changes in the amplitude or shape of the T wave. TWA has been considered an important non-invasive indicator for detecting sudden cardiac death in addition to being included in modern medical equipment. HRV is a SCD risk stratification index that allows to measure time variation between consecutive heartbeat sets. Considering these indices and adding digital signal processing techniques as dictionary learning that can be found in the breakdown of signals on a specified basis, for instance, the Fourier transform, a hybrid method has been developed. The proposed method identifies the main characteristics of ECG signal by obtaining a sparse representation that adapts a matrix (dictionary) in order to use it for highlighting the TWA characteristics and then use them for detecting SCD. Experimental results show an improvement of 32% compared to the Physionet TWAnalyser program by using synthetic data set and an improvement of 20% over public databases. This research presents an innovative strategy for predicting sudden cardiac death using TWA analysis and dictionary learning. To evaluate the proposed methodology, public databases were used and synthetic signals were generated.