HEART DISEASE PREDICTION USING MACHINE LEARNING

Abstract

In the ongoing battle against heart disease, where precision is paramount, the realm of machine learning emerges as a potent ally, harnessing predictive capabilities derived from accumulated knowledge. This research delves into the complexities of heart disease, employing an array of supervised learning algorithms ranging from Logistic Regression to Random Forest and Gradient Boosting Classifier. Leveraging the Cleveland dataset from the UCI repository, the study focuses on 303 patient instances and 76 attributes, narrowing down to 14 key attributes for a concentrated evaluation of algorithm performance. The overarching goal is to unravel the data's secrets, predicting the likelihood of future heart disease in patients. A surprising revelation unfolds as k-Nearest Neighbours (KNN), celebrated for its simplicity and efficacy, emerges as the top-performing classifier with the highest accuracy score. This discovery paves the way for exciting prospects in heart disease diagnosis, suggesting that further refinement and optimization of the KNN model could lead to earlier and more precise predictions. This breakthrough has the potential to empower healthcare professionals to implement timely preventive measures, tailor treatment plans, and ultimately contribute to saving lives.