DEVELOPMENT OF A MACHINE LEARNING-BASED HEART DISEASE DETECTION SYSTEM FOR EARLY DIAGNOSIS AND CLINICAL DECISION-MAKING IN RESOURCE-CONSTRAINED RURAL SETTINGS

Heart disease remains a major cause of morbidity and mortality worldwide and is an emerging public health challenge in rural Nigerian communities, where access to diagnostic facilities and specialized healthcare services is limited. Therefore, this study presents the development of a machine learning–based heart disease detection system aimed at improving early diagnosis and clinical decision-making in resource-constrained rural settings. In the study, an Artificial Neural Network (ANN) model was designed and implemented using the Extreme Programming (XP) methodology to ensure flexibility, rapid iteration and continuous system validation. The model was trained on a hybrid dataset comprising primary health records collected from rural health centres in Enugu State, Nigeria and a secondary publicly available heart disease dataset from Kaggle, yielding a total of 20,428 samples. Comprehensive data pre-processing techniques including imputation, normalization, feature selection, feature transformation, and class balancing using SMOTE were applied to enhance model performance. The ANN architecture employed multiple hidden layers with ReLU activation, dropout regularization and a sigmoid-based output layer for binary classification. Experimental results from the study implementation demonstrated a training accuracy of 90.50% and a test accuracy of 89.10%, with strong precision, recall, and F1-score values, indicating reliable generalization and effective heart disease prediction. Comparative analysis with existing state-of-the-art techniques further validated the robustness of the proposed approach. The developed system provides a scalable and practical decision-support tool capable of assisting healthcare workers in early heart disease detection, thereby improving patient outcomes and reducing mortality in underserved rural communities.