Prediction of Sudden Cardiac Death (SCD) Using Time-Frequency Analysis of ECG Signals

Despite the significant decline in coronary artery disease (CAD) mortality in the second half of the 20th century، sudden cardiac death (SCD) continues to claim 250 000 to 300 000 US lives annually. Even in the presence of advanced first responder systems for resuscitation of out-of-hospital cardiac arrest، the overall survival rate in a recent North American analysis was 4. 6%. If there are existed suitable ways to predict sudden cardiac death، doctors can make better decisions for patients at risk. In this paper، we investigate a way to predict sudden cardiac death. To do this، after the extraction of the HRV signal from ECG signal، some nonlinear and time-frequency features have been extracted from HRV signal. Then، the dimension of the feature space is reduced by applying the feature selection and PCA. Finally، healthy people and people at risk of SCD are classified using an MLP neural network. To evaluate the capabilities of analytical methods in classification، we have compared the classification rates for nonlinear and TF features، separately and in combination. The results show that there are features in the HRV signal of SCD patients just near the occurrence of SCD، which is quite different from normal people. Also، results show that the combination of time-frequency and nonlinear features have a greater ability to detect this difference. It has also been investigated that there are precious information in four minutes before the incident of SCD to predict the death; and this is enough time to save the patient by doctors or medical centers.