Secure and Privacy-Preserving IoT-Based Arrhythmia Detection with Federated Learning

Arrhythmias pose a significant challenge for early diagnosis and treatment, necessitating advanced detection techniques. This paper introduces the Secure and PrivacyPreserving IoT-Based Arrhythmia Detection (SPIoT-AD) model, a federated learning (FL) based approach designed to enhance arrhythmia detection from ECG signals within IoT healthcare frameworks. Utilizing a secure multi-layered architecture that processes data at the edge, the methodology ensures data privacy and real-time monitoring, adapting to individual patient profiles. The model showcases exceptional performance metrics, demonstrating an accuracy of 99. 85 %, sensitivity of 98. 84 %, specificity of 99. 93 % and with a precision of 98.63% in the classification of multiple arrhythmia cases. Further, it also demonstrates a highest classification accuracy of 99.97% in the detection of normal cases. These results demonstrate the model’s potential as a reliable tool for arrhythmia detection and classification, setting a foundation for future integration into clinical workflows, ensuring security and privacy of patient data.