Improving Heart Disease Prediction Accuracy through Machine Learning Algorithms

Abstract

This study explores the application of a range of machine learning and deep learning techniques for predicting cardiovascular diseases. Various models, including Random Forest, Logistic Regression, Gradient Boosting, AdaBoost, Support Vector Machine (SVM), XGBoost, and both Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks are evaluated. A comprehensive evaluation is conducted by considering supplementary metrics, refining hyperparameter tuning, assessing feature importance using SHAP, comparing traditional machine learning with deep learning approaches, and examining the clinical relevance. It concludes that XGBoost achieves the highest accuracy (88%), and notes that CNN and LSTM may prove beneficial with larger datasets. Moreover, the study investigates the practical applications of these models, focusing on their potential integration into clinical decision support systems.