Edge AI Architecture for Real-Time Arrhythmia Detection Using Wearable ECG Devices
Abstract
As the world's population continues to age rapidly, more sophisticated IoT (Internet of Things) frameworks and smart home health assistance technology have emerged to help people remain independent while living at home. While strides have been made in the development of smart homes and robotic assistants that make it easier to monitor individuals in their homes, there are still limitations when it comes to constantly tracking people's body functions through traditional means of monitoring their cardiovascular system using new wireless medical technologies. The use of a centralized cloud computing architecture to process ECG data continuously creates a number of problems for monitoring cardiovascular conditions using on-device ECG devices. Therefore, this research presents an Edge AI architecture specifically for real-time arrhythmia identification using wearable ECG devices. Thus, this architecture reduces the latency associated with dying of a heart rhythm disorder to only milliseconds from the time it takes for the ECG device to send the ECG data to the centralized cloud server for processing, yet provides real-time detection of abnormal heart events while dramatically reducing power consumption and protecting user privacy. In addition to providing a complete health monitoring solution for those aging in place, the proposed Edge AI architecture can be easily incorporated into existing multi-agent smart home and Care Bot architectures.