Arrhythmia classification based on wavelet transformation and random forests

Cardiovascular disease accompanied by arrhythmia reduces an individual’s lifespan and health, and long term ECG monitoring would generate large amounts of data. Fortunately, arrhythmia classification assisted by computer science would greatly improve the efficiency of doctors’ diagnoses. However, due to individual differences, noise affecting the signal, the great variety of arrhythmias, and heavy computing workload, it is difficult to implement these advanced techniques for clinical context analysis. Thus, this paper proposes a comprehensive approach based on discrete wavelet and random forest techniques for arrhythmia classification. Specifically, discrete wavelet transformation is used to remove high-frequency noise and baseline drift, while discrete wavelet transformation, autocorrelation, principal component analysis, variances and other mathematical methods are used to extract frequency-domain features, time-domain features and morphology features. Furthermore, an arrhythmia classification system is developed, and its availability is verified that the proposed scheme can significantly be used for guidance and reference in clinical arrhythmia automatic classification.