CARDIO TWIN-H: AN AI-INTEGRATED REAL TIME CARDIOVASCULAR DIGITAL TWIN SYSTEM FOR PROACTIVE RISK REDICTION AND CLINICAL DECISION SUPPORT

Abstract

Cardiovascular diseases are still the topmost killers of people all around the world, claiming approximately 17.9 million lives each year. Traditional tools to assess the risk of developing cardiovascular diseases, including the Framingham Risk Score, do not incorporate interactivity, explainability, and real-time simulative capabilities that would enable doctors to provide proactive and personalized care. In this work, we present CardioTwin-H, an AI-Integrated Real-Time Cardiovascular Digital Twin System with Hardware Sensor Integration. CardioTwin-H is a combination of a MAX30102 pulse oximetry and heart rate sensor connected to a Raspberry Pi 4 microcomputer and a machine learning-based platform. Data are streamed to a FastAPI backend server, where an ensemble of Random Forest, Gradient Boosting, and Logistic Regression algorithms calculates the cardiovascular risk score in real time. Reports based on the SHapley Additive Explanations framework are used to interpret data and provide clinically meaningful explanations. Finally, a scenario simulation engine allows for projecting the influence of clinical intervention on the predicted health state of patients. Overall, the system can be deployed at hardware costs of less than six thousand Indian Rupees.