| dc.description.abstract | Coronary heart disease (CHD) remains one of the leading causes of death worldwide and poses a major challenge for healthcare systems, particularly in developing countries such as Indonesia. This study aims to develop an early risk prediction model for CHD using a machine learning–based approach through hybrid integration of Neural Network (Multi-Layer Perceptron/MLP) and Random Forest (RF) to enhance both accuracy and interpretability. The dataset consists of 1,000 patient records containing demographic and clinical attributes, including age, blood pressure, cholesterol levels, smoking history, diabetes, and hypertension. The research process includes data preprocessing (imputation, outlier detection using the IQR method, and normalization with MinMaxScaler), feature engineering, and dimensionality reduction using Principal Component Analysis (PCA). The models were evaluated using Accuracy, Precision, Recall, F1-Score, and ROC-AUC metrics across three approaches: MLP, RF, and hybrid integration. The results show that the hybrid model achieved the best performance, with an Accuracy of 0.9958 and ROC-AUC of 0.9991, outperforming the individual models. These findings demonstrate that integrating the two algorithms produces a more stable, accurate, and clinically relevant predictive system, thereby supporting early detection of CHD in a more effective and human-centered manner. | en_US |
