Congestive Heart Failure from ECG Prediction Using Empirical wavelets transform Algorithm

Subject Areas : Communication

Nazanin Tataei Sarshar ¹ , Mehdi Abdossalehi ²

1 - Department of Electrical Engineering, North Tehran Branch, Islamic Azad university, Tehran, Iran.
2 - Department of Engineering, Islamshahr Branch, Islamic Azad University, Islamshahr, Iran

Received: 2021-06-28 Accepted : 2021-10-17 Published : 2022-03-01

Keywords: Heart failure, Heart congestion, Empirical wavelets transform,

Abstract :

Heart diseases are the leading cause of death worldwide as such the use of an advanced information processing method to diagnose heart disease is one of the most critical fields of medical research. Among various heart diseases, congestive heart failure refers to a difficulty in the heart's pumping, and its symptoms vary depending on the body organ being the most involved in reducing pumping performance. Accordingly, several studies have been conducted to diagnose heart problems using Heart rate variability (HRV) analysis of cardiac signals even though the signals are not accurate enough. The HRV signal extracted from the ECG signal was analyzed to classify the Congestive heart failure’s signals and normal signals in the proposed method. Signal decomposition into a series of subbands was performed using Empirical Wavelets Transform (EWT), and the values ​​were calculated based on different subbands. In this case, the extracted features were classified by the SVM classification method. The method was used to classify normal individuals and those with CHF into two normal or abnormal groups. Finally, by implementing the proposed model and simulating the data of the PhysioNet site, the CHF problem could be automatically detected. The evaluation of the proposed method in comparison to other methods revealed that the proposed method has a significant advantage over other methods, as indicated by the accuracy value of 98.30.Heart diseases are the leading cause of death worldwide as such the use of an advanced information processing method to diagnose heart disease is one of the most critical fields of medical research. Among various heart diseases, congestive heart failure refers to a difficulty in the heart's pumping, and its symptoms vary depending on the body organ being the most involved in reducing pumping performance. Accordingly, several studies have been conducted to diagnose heart problems using Heart rate variability (HRV) analysis of cardiac signals even though the signals are not accurate enough. The HRV signal extracted from the ECG signal was analyzed to classify the Congestive heart failure’s signals and normal signals in the proposed method. Signal decomposition into a series of subbands was performed using Empirical Wavelets Transform (EWT), and the values ​​were calculated based on different subbands. In this case, the extracted features were classified by the SVM classification method. The method was used to classify normal individuals and those with CHF into two normal or abnormal groups. Finally, by implementing the proposed model and simulating the data of the PhysioNet site, the CHF problem could be automatically detected. The evaluation of the proposed method in comparison to other methods revealed that the proposed method has a significant advantage over other methods, as indicated by the accuracy value of 98.30.