Improved Stochastic Gradient Descent-Decision Tree (ISGD-DT) Framework for Intelligent Heart Disease Prediction

Abstract

This research presents an innovative system architecture for heart disease prediction that integrates Improved Stochastic Gradient Descent (ISGD) with a Decision Tree (DT) classifier. The ISGD-DT model addresses challenges in existing predictive models, such as imbalanced datasets, limited generalizability, and suboptimal accuracy, by leveraging hierarchical layers, graph databases, and decision trees for robust classification outcomes. Validated using benchmark datasets from the UCI Machine Learning Repository, including the Cleveland and Hungarian heart disease datasets, the model demonstrates superior performance with accuracy rates of 93.17%, 88.39%, and 96.29% across different datasets. These results highlight the model's reliability and robustness, making it a valuable tool for improving predictive modeling in healthcare. This research underscores the potential of combining advanced optimization techniques and classification algorithms to enhance the accuracy and applicability of medical prognostics.