Prediksi Diagnosa Penyakit Jantung (Cardiovascular Diseases) Menggunakan Algoritma Machine Learning

Abstract

Heart disease remains a global health concern, being the leading cause of mortality with substantial impacts on the population. This research addresses the challenges in early detection and prediction of heart diseases, considering the complex and diverse nature of Cardiovascular Diseases (CVD). With limitations in diagnostic tools and healthcare resources, the study explores the application of machine learning algorithms for accurate predictions. Building upon previous research, various machine learning algorithms, including Random Forest, Multilayer Perceptron, Gaussian Processes, and M5P, were employed to predict heart disease-related data. The research involved comprehensive data pre-processing, visualization, model fitting, and evaluation stages. The dataset, sourced from the Hungarian Institute of Cardiology, comprised 14 attributes. Results demonstrated the effectiveness of the selected machine learning models, with Random Forest exhibiting outstanding performance, closely followed by Multilayer Perceptron. Gaussian Processes performed relatively well, while M5P provided a complex model structure offering additional insights. The use of 10-fold cross-validation enhanced the stability of model evaluation. Statistical analysis and data visualization contributed to a thorough understanding of model performance and dataset characteristics. In conclusion, this research contributes to developing accurate predictive models for heart disease detection. The findings offer valuable insights into algorithm performance and dataset characteristics, guiding future health science and information technology efforts for improved preventive and diagnostic measures. The methodology employed, including machine learning algorithms and cross-validation, presents a robust approach for future research in cardiovascular health prediction