A Machine Learning-Based System for Medical Assistance in the Diagnosis of Cardiovascular Diseases

This paper investigates the early diagnosis of cardiovascular diseases (CVD) through machine learning (ML), utilizing the database from the University of California, Irvine (UCI). This paper evaluates ten ML models. Among the evaluated ML models, the StackingCVClassifier, integrating XGBoost, knearest neighbors (KNN), and support vector machine (SVM), stood out with the highest accuracy of 92.64%. Moreover, to prioritize accessibility for cities with little infrastructure, a scale was developed to evaluate the difficulty of collecting attributes and adapting the training of models using the database with and without hard-to-collect attributes. Finally, a user interface was developed to enable the system for health professionals.