A Comparative Analysis of Traditional and Neural Network-Based Models for Heart Disease Prediction

ABSTRACTHeart disease is the most prevalent cause of morbidity and mortality all overthe world hence the importance of precise and timely means of predicting thedisease. The research paper provides comparative analysis of machine learningalgorithms-Support Vector Machine (SVM), k- Nearest Neighbor (KNN) andArtificial Neural Networks ( ANN ) to build an early detection model of heartdisease on the basis of some of the most vital clinical indicators, i.e. type ofchest pain, blood pressure, cholesterol level and electrocardiographic findings.The proposed framework aims to identify individuals at elevated cardiovascularrisk, thereby enabling earlier intervention and reducing adverse outcomes.Initial models developed using SVM and KNN demonstrated improvedpredictive performance over traditional classifiers such as Na ̈ıve Bayes.However, to address the limitations of linear decision bound- aries and locallearning, an ANN-based model was introduced to capture com- plex nonlinearrelationships within the data. The comparison of the experimental resultsshowed that ANN has a better performance than the other models in manyaspects, such as precision of prediction, recall or true positive rate, and f1-score as well as area under a ROC curve, and ANN is robust and reliable asdiagnostic support. The findings suggest that incorporating deep learning intoclinical decision-making tools can significantly enhance predictive accuracy andcontribute to improved patient outcomes in cardiovascular care.