Heartify: A Federated Learning-Based Application for Early Heart Disease Detection

Cardiovascular diseases (CVDs) are the leading cause of global mortality, making early detection critical for timely intervention. Traditional machine learning-based detection systems often compromise patient privacy by requiring centralized data storage. This study presents Heartify, a privacy-preserving heart disease monitoring application that leverages Federated Learning (FL) to train models across distributed smart devices. By integrating Federated Averaging (FedAvg) and Federated Matched Averaging (FedMA), the proposed system optimizes decentralized model updates while ensuring compliance with GDPR and HIPAA regulations. The results demonstrate that FedMA achieves an accuracy of 85.32 %, surpassing traditional deep learning models. This research highlights FL’s potential to enable scalable, real-time, and privacy-preserving heart disease prediction across diverse wearable devices.