An Ontology-Based Approach for Heart Disease Prediction Leveraging Decision Trees in Healthcare

Abstract An ontology-driven system for the diagnosis of heart disease combines data-driven methods with semantic reasoning. The dataset consists of major healthcare attributes like patient information, clinical findings, and diagnostic indicators. This information is processed in a methodical way into a clinical ontology, which facilitates structured, semantically enhanced representation of medical entities and their relationships. The system uses a hybrid strategy based on decision trees and Graph Neural Networks (GNNs) to enhance predictive performance. Decision trees offer interpretability through the identification of key decision paths, whereas GNNs utilize connected clinical ontology for richer relational analysis. Semantic Web Rule Language (SWRL) rules also enrich reasoning by deducing new knowledge from implicit relationships and patterns. The rules combine risk factors such as hypertension, cholesterol, and lifestyle to predict heart disease with high accuracy. The integration of rule-based reasoning and machine learning provides a complete and context-sensitive system. Performance is measured using accuracy, precision, recall, F1-score, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC) to provide solid diagnostic evaluation. The hybrid model of decision trees, GNNs, and ontology-based reasoning with boosting via Semantic Web Rule Language (SWRL) rules shows great predictive accuracy and decision support improvements. The clinical ontology provides interpretability, flexibility, and scalability for future data sets and changing medical knowledge. This work emphasizes the opportunity of combining ontology engineering, machine learning, and semantic reasoning in enhancing heart disease prediction and diagnosis. The system assists healthcare workers in making well-informed decisions, improving diagnosis accuracy, and maximizing patient management.