9th International Conference on ICT for Sustainable Development

Authors - Sonam Nagar, Karan Verma, Sachin Singh, Kumar Shashvat
Abstract - Every year, the National Institutes of Health (NIH) and the American Heart Association (AHA) collaborate to disseminate the most recent data on heart disease, stroke, and cardiovascular risk factors. These include basic health behaviors such as physical inactivity, diet, sleep habits, smoking, and obesity as well as health factors including blood pressure(bp), cholesterol, glucose regulation and metabolic syndrome, all of which have a substantial impact on cardiovascular health. Globally, these ailments account for 31% of all deaths and stand as the leading cause of mortality. Projections suggest that by 2030, the death toll from cardiovascular disease could climb to 22 million individuals. Presently, American Heart Association data indicates that half of all adults in the United States grapple with some form of cardiovascular disease. This paper conducts a comparative analysis of three classifiers to predict heart disease cases while minimizing the number of attributes required for accurate classification. We utilize a publicly accessible dataset for predicting heart failure, employing three supervised machine learning classification algorithms: SVC, DT, and RF. Achieving a significant accuracy of 90.21% with SVC on the public dataset is noteworthy. Additionally, Random Forest achieved 88.40%, while Decision Tree yielded 78.62%. Our study highlights the importance of employing different classifiers and underscores the advantages of employing a robust feature selection method for predicting heart disease with minimal attribute usage instead of considering all available features. In conclusion, the Support Vector Classifier (SVC) emerged as the most effective model for predicting heart disease in our experiment.