ECG Classification using Kernel Extreme Learning Machine
Sahil Dalal, Virendra Prasad Vishwakarma, Varsha Sisaudia · 2018
Almost all humans have different ECG waveforms though the frequency range is quite narrow which is confined within few hundred hertz only. The doctors, especially cardiologist use Electro-Cardio-Graphy (ECG) to acquire first-hand knowledge about the well-being of human heart. Generally, ECG provides indicative information for most of the cardiac ailments. ECG can also be used effectively for other non-medical applications such as biometrics, person identification etc., but in such cases large amount of ECG data is required to be processed and classified to obtain desired results. This process is quite complicated, tedious and time consuming. The proposed method classifies UCI repository dataset of ECG signals based upon the parameters that include P, Q, R, S, T peaks, their envelopes width etc. This method is a comprehensive measure of non-linearity and gives better results in terms of time complexity. It provides a quality detection technique in comparison to the other methods, used earlier in this field.