| Autor: |
Singhal, Amit, Agarwal, Megha |
| Zdroj: |
Multimedia Tools & Applications; Mar2024, Vol. 83 Issue 9, p27243-27258, 16p |
| Abstrakt: |
Malfunctioning of the electrical system of heart causes irregular heart rhythms, which may result in sudden cardiac death (SCD). An early detection of SCD risk can help in saving many lives across the globe. Non-invasive diagnostics like electrocardiogram (ECG) are being actively explored to address this problem. In this work, we build an efficient system to predict SCD risk using ECG signals, at least 10 minutes before the ventricular fibrillation (VF) onset, allowing sufficient time for taking preventive action. The signal is de-noised and decomposed into sub-band components using a Fourier-based decomposition technique. Local variations in the signal samples are extracted using look ahead pattern (LAP) and represented in the form of histogram. Principal component analysis (PCA) is applied and only the first five features are passed to different machine learning classifiers to detect the SCD risk. The proposed method achieves 100% accuracy in identifying SCD cases from non-SCD cases, which include congestive heart failure (CHF) and normal sinus rhythm (NSR). Further, the performance of the algorithm is analyzed in noisy conditions, considering different lengths of the proposed feature. The performance analysis highlights the strength of the proposed method for an efficient implementation in real-time systems. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink