Applying Machine Learning Classifiers on ECG Dataset for Predicting Heart Disease

Sudden demise from heart disease is rising in a terrible rate and this disease has become a common cause of death worldwide. But it is a matter of hope that heart diseases are avertible by making simple lifestyle changes coupled with early prognosis which can greatly improve its recovery. Identifying high risk patients is difficult due to the multifaceted characteristic of various threat factors such as high cholesterol, high blood pressure, diabetes etc. Most of the time, diagnosis of heart disease depends on doctor’s observation and expertise instead of utilizing the large amount of knowledge-rich medical dataset. To change the situation, scientists and doctors have turned to machine learning techniques to evaluate screening results along with other medical parameters to predict heart disease. For heart disease prediction, this study implements five machine learning algorithms including Support Vector Machine, Logistic Regression, K-nearest Neighbor, Naive Bayes, and Ensemble Voting Classifier on a dataset with 1190 records accumulated from UCI repository. The dataset combines five independent ECG dataset which gives us an extra edge to achieve our objectives. Relation among the attributes in the dataset is analyzed before the accuracy is calculated. Among the five classification algorithms, Support Vector Machine outperforms other classifiers with the accuracy of 85.49%. We hope this study will ensure early diagnosis of heart disease and increase the chance of survival.