Classification of heart rate variability signals using higher order spectra and neural networks

In this paper, the heart rate variability signals were utilized to automatically discriminate between subjects with normal sinus rhythm and patients with low heart rate variability such as those suffering from congestive heart failure (CHF) and myocardial infarction (MI) diseases. Traditional techniques and Higher Order Spectral analysis (HOS) were used to extract the main features from the HRV signals. Also, higher order spectra (HOS) estimators are the sum of the estimated bispectrum, bicoherence index and normalized bispectral entropy. An Artificial neural network classifier (ANN) was proposed to compare the classifier performance for automatically classifying the aforementioned diseases. Results have shown that using HOS parameters give high rates for classifying heart diseases. Classification rate reaches to 98.78%.