CARDIOAI: An Intelligent Heart Disease Prediction System Using Machine Learning

Abstract

Heart disease still ranks among the leading causes of mortality; therefore, efficient and easily available screening is crucial for preventing deaths related to cardiovascular issues. This paper describes CardioAi – a web-based platform, which analyzes the clinical features of a patient and assesses the cardiovascular risk using machine learning. Four supervised models – Logistic Regression, Random Forest, Support Vector Machine and Decision Tree are applied to build the model based on patients’ medical records, while the final prediction is provided along with the risk visualization, PDF report creation and advisory tool working on the basis of rules instead of being presented as a number. The system is built as a full-stack web application using React.js for front-end development, FastAPI for back-end development and MySQL for data storage. Two different user roles are considered: Patient and Doctor, whose access is regulated via role-based authentication allowing both a patient and their physician to monitor the history of predictions. On the evaluation set consisting of more than 50,000 patient records, Random Forest classifier demonstrates the highest accuracy among four models – approximately 88% versus Support Vector Machine (~86%), Logistic Regression (~85%) and Decision Tree (~82%).