Optimizing the Early Detection of Heart Disease with Improved Feature Selection Using Machine Learning Model

Cardiovascular disease(CVD) deaths has increased from 12.1 million in 1990 to 20.5 million in 2021. Traditional machine learning methods for CVD detection are often chal-lenging to select optimal features. This paper mainly focuses on The ECG signal Preprocessing, feature extraction and ECG beat classification phases comprise the three stages of this methodology. To extract the wavelet coefficients and to de-noise the signal, one uses the Discrete Wavelet Transform. Each ECG beat's characteristics are provided as inputs to the classifier. A dataset comprising approximately 2000 The ECG signals are taken from the MIT-BIH Arrhythmia Database. These signals undergo preprocessing to ensure uniformity and readiness for analysis. Subsequently, feature extraction is performed to distill relevant information from the signals. For classification purposes, two distinct algorithms [1] are implemented named as Support Vector Machine (SVM) and Decision Tree are compared with Depth wise Convolutional Neural Network(CNN), Each Algorithm receives the preprocessed signals as input, which are then subjected to feature extraction. Following this, the classifiers are trained and tested on the extracted features [2]–[4] to discern their effectiveness in accurately classifying the ECG signals.