A streamlit-powered cloud platform for machine learning-driven early detection of cardiovascular diseases

Abstract

Cardiovascular diseases (CVDs) are a major contributor to global morbidity and mortality, highlighting the need for early detection and prevention. This study introduces CardioPredict AI, a cloud-based system using advanced machine learning (ML) for CVD prediction. It offers scalable, accessible, and real-time diagnosis. The system leverages a comprehensive patient dataset that integrates multiple clinical features, including age, cholesterol levels, and blood pressure. Data preprocessing involved imputation, normalization, one-hot encoding, and the selection of 12 key features. The random forest model achieved an accuracy of 90.21%, a recall of 94.75%, and an F1-score of 91.31%, meeting the medical standards for heart disease prediction (recall >90%; false negatives <20). Cross-validation yielded a recall of 0.8940 ± 0.0889. Key features include personalized recommendations, real-time risk assessment through a Streamlit application, SHapley Additive exPlanation-based interpretability, and a dashboard for patient metrics. This study highlights the potential of ML and cloud computing to reduce the burden of CVDs through early detection.

Ähnliche Arbeiten