Machine Learning in Heart Disease Prediction

A Comparative Study of Algorithms

Abstract

Cardiovascular diseases (CVDs), simply heart diseases, are the leading cause of death globally, needing effective early detection tools. Traditional methods for diagnosis are time-consuming and they cost too much. The goal of this project is to conduct a comprehensive comparison analysis on the applicability of machine learning algorithms for predicting heart diseases. We used a publicly available dataset. We conducted Exploratory Data Analysis (EDA) and data preprocessing, including feature scaling and encoding, to make the data suitable for machine learning algorithms. The study included utilizing fundamental machine learning methods like Logistic Regression and Support Vector Machines, as well as more advanced techniques such as Artificial Neural Networks. The performance of our classifiers was evaluated using the metrics recall, precision, F1 Score, and accuracy. Based on our study, Logistic Regression was the most accurate model, with the accuracy rate of 90%. Furthermore, we found out that hyperparameter tuning and data standardization increased classifier performance. Our findings provide a thorough guidance for healthcare practitioners and data scientists interested in using machine learning for heart disease prediction, including all steps from data preparation through model evaluation.