A Clinical Support System for Prediction of Heart Disease using Ensemble Learning Techniques

In our world heart disease is one of the major health concerns for public and is the reason of a sizable number of fatalities each year. According to the WHO, over 17 million people died from cardiovascular disease (CVD) in 2019. Effective prevention and treatment of heart disease depend on early detection, and machine learning algorithms have demonstrated great promise in predicting chance of heart disease causes by various unhealthy factors. Development of machine learning and deep learning based techniques have made easy to detect the heart diseases. These models utilize a combination of demographic, clinical, and lifestyle factors, such as age, sex (M/F/others), Blood Pressure (BP), cholesterol levels, smoking habit status, and family's heart disease history, among others. This paper analyzes the performance over selection of features and gives the best set of features for selected algorithm. S tacking ensemble learning technique and K fold validation was used in this paper and predictions were made using Decision Tree, KNN and SVM. Before building the models, preprocessing of data and feature selection were completed. To calculate our model's performance, we have included the following parameters like precision, recall, accuracy. The Random Forest (RF) machine learning performed the best with given dataset, with an accuracy score of 99.02%.