HEART DISEASE PREDICTION USING EXTREME LEARNING MACHINE AND HEART SOUND FEATURE ANALYSIS
Keywords:
Disease Prediction, Phonocardiogram (PCG), Heart Sound Classification, Machine Learning, Extreme Learning Machine (ELM), Support Vector Machine (SVM), Feature Extraction, Mel-Frequency Cepstral Coefficients (MFCC).Abstract
One of the major reasons of death in today’s world is due to the diseases related to the heart and there is a need to come up with new methods for proper diagnosis of such diseases. The prediction of these cardiovascular pathologies is based on the analysis of the heart sounds using microphones during their cardiac cycle. The ELM and SVM algorithms have been used to check the efficiency of categorizing these heart sounds into classes such as Normal, Murmur, Extrasystole, Artifact, and Extrahls based on feature analysis. In addition, for the classification process, some additional and advanced feature extraction techniques such as Mel-Frequency Cepstrum Coefficients (MFCCs), FFT, Continuous Wavelet Transform (CWT), and Discrete Wavelet Transform (DWT) have been used. The results provided in this paper show that ELM is a more advanced technique for diagnosing or suspecting possible heart problems as opposed to the use of SVM techniques that have been used in gyro type systems. It will be imperative to undertake further research projects that incorporate health monitoring in smart devices using component-based cloud computing and deep learning.
