Medical Decision Support System for Diagnosis of Cardiovascular Diseases using DWT and k-NN

Heart disease is a cardiovascular disorder that is most widespread cause of death in many countries all over the world. In this work, k-Nearest Neighbor machine learning tool was used to classify Electrocardiography (ECG) signals and satisfactory accuracy rate was achieved in classification of ECG signals. The model automatically classifies the ECG signals into 5 different kinds: normal, Premature Ventricular Complex (PVC), Atrial Premature Contraction (APC), Right Bundle Branch Block (RBBB) and Left Bundle Branch Block (RBBB). The best averaged performance over randomized percentage-split is also obtained by k-Nearest Neighbor (k-NN) classification model. Some conclusions concerning the impacts of features on the ECG signal classification were obtained through analysis of different parameters of kNN. The analysis suggests that kNN modeling is satisfactory performances in at least three points: high recognition rate, insensitivity to overtraining and computational time it takes for classification. The combined model with DWT and k-NN achieves the good. Obtained result shows that the suggested model have the potential to obtain a reliable classification of ECG signals, and to support the clinicians for making an accurate diagnosis of cardiovascular disorders. Keywords: Electrocardiogram (ECG); Discrete Wavelet Transform (DWT); k-Nearest Neighbor (k-NN); Heart Arrhythmia; Premature Ventricular Complex (PVC); Atrial Premature Contraction (APC); Right Bundle Branch Block (RBBB); Left Bundle Branch Block (RBBB)