Secure Decentralized ECG Prediction: Balancing Privacy, Performance, and Heterogeneity
Bagesh Kumar, Sohan Kumar, Yash Vikram Singh Rathore, Akash Raj, Vanshika Singh Andotra, Rishik Gupta, Prakhar Shukla · 2025
In recent years, there has been a surge in utilizing generative artificial intelligence (AI), particularly generative adversarial networks (GANs), to enhance decentralized prediction models for analyzing electrocardiogram (ECG) data. This book chapter explores how generative AI methods address limitations of conventional techniques in ECG analysis, such as data scarcity and privacy concerns associated with centralized storage. It discusses the technical aspects of generative AI, including generating synthetic ECG signals, improving data quality, detecting anomalies, and creating realistic clinical scenarios. In addition, the chapter examines the implementation of decentralized prediction models using federated learning frameworks to ensure data privacy while enabling collaborative training across distributed ECG datasets. This advancement in generative AI has the potential to revolutionize ECG data processing and predictive modeling in decentralized healthcare systems, enabling personalized treatment plans and remote monitoring, thereby fostering a new era of patient-centric care delivery.