Abstract:
Cardiovascular diseases are one of the most extensive forms of deadly disease all over the world; thus, early identification of cardiac issues is vital to ensure fatal outcomes are minimized and treatment is improved. The objective of this study is to assess the effectiveness of various machine learning techniques at predicting heart disease diagnoses from refined clinical and lifestyle data features. We then used six classification models (XGBoost, Decision Tree, Random Forest, Support Vector Machine, Naive Bayes, and Multi-Layer Perceptron Neural Network) to validate our dataset. The dataset was well preprocessed as it included missing value, normalisation, encoding and feature selection techniques with high data integrityatges. The model’s effectiveness was evaluated using accuracy, recall, precision, F1-score, ROC AUC, along with calibration curves and SHAP- based interpretability. XGBoost was the highest performing model with a 96.18% accuracy and it achieved the best one (the best balance between recall and precision), whilst Decision Tree and Random Forest also produced reasonable balanced scores. In conclusion, the results indicate that tree-based fusion methods are a great compromise between accuracy and interpretability, and viable for clinical implementation. These findings demonstrate the potential and challenges of applying AI to cardiac diagnostics while considering the ethical, societal, and sustainability issues central to real-world healthcare implementation.
