Machine Learning Approaches for Coronary Artery Disease Detection: A Comprehensive Technical Investigation

Abstract

This study aims to analyze and predict heart disease presence using the Cleveland Hungarian Switzerland dataset, employing a robust methodology encompassing data collection, pre-processing, exploratory data analysis (EDA), data splitting, and machine learning modelling. The dataset, comprising 1025 rows and 14 columns, provides a rich foundation for investigating cardiovascular health factors. The pre-processing phase involves identifying and handling missing values, followed by converting the dataset to the 'int64' data type. EDA employs visualizations such as count plots, histograms, and correlation matrices to uncover patterns and relationships within the dataset.The data splitting step partitions the dataset into training and testing sets, essential for assessing the Random Forest Classifier's performance. This machine learning model, chosen for its adaptability and efficiency, demonstrates exceptional predictive accuracy of 98.53%, with high precision, recall, and F-score. Comparative analysis with existing models reveals the proposed Random Forest model's superior performance, showcasing the potential for customization to enhance predictive accuracy. The study contributes valuable insights into cardiovascular health analysis and predictive modeling, emphasizing the importance of comprehensive methodologies for effective exploration and prediction in medical datasets.