Improving the diagnosis of sudden cardiac death by using non-linear features of the ECG signal and hybrid RBF

Sudden cardiac death (SCD) leads to killing millions of people worldwide every year. In this article, sudden cardiac death is predicted by utilizing electrocardiogram signal processing. For this purpose, after extracting the signal of heart rate variations from the electrocardiogram signal, temporal and non-linear characteristics have been extracted. In the next step, by applying LDA to the combined feature vector, the feature dimensions are reduced and finally, healthy people and high-risk people are classified through Hybrid-RBF classifiers. The obtained results show that there are features in the signal of heart rate variations related to risk-taking individuals near the occurrence of sudden cardiac death, which completely distinguish them from healthy persons. It has also been shown that from 6 minutes before the occurrence of cardiac death, this increase in the probability of risk is quite evident, so that as we get closer to the occurrence of the accident, the probability of its occurrence also increases, and this is enough time to adopt strategies to prevent it. The simulation results achieved by the data available in the MIT-BIH database prove the ability of the presented methods in accurate diagnosis.

Improving the diagnosis of sudden cardiac death by using non-linear features of the ECG signal and hybrid RBF Keywords:

variations in heart rate , sudden cardiac death , linear features , nonlinear features