Enhancing Predictive Accuracy in Heart Disease Detection via Hyperparameter-Tuned KNN and Grid Search CV

Abstract

Cardiovascular diseases (CVDs) are the leading cause of global mortality. Accurate and early risk prediction is crucial for effective prevention and treatment. This thesis explores the application of advanced machine learning techniques to identify patterns and risk factors in heart disease, aiming to improve upon traditional statistical methods. A comprehensive methodology was employed, including data preprocessing, feature analysis, and rigorous hyperparameter tuning of various models. The research demonstrates that well-tuned machine learning models, particularly ensemble methods, can achieve superior predictive accuracy compared to conventional approaches. The findings highlight the most influential clinical features contributing to heart disease risk, providing valuable, data-driven insights. However, the study acknowledges several limitations, including the challenge of generalizability due to the use of a single, static dataset, and the inherent lack of interpretability in complex "black box" models. The research concludes that while machine learning holds immense promise for personalized medicine, future work is needed to develop models that are not only accurate but also transparent and validated on diverse, realworld data to facilitate their adoption in clinical practice.