Real-time health monitoring system based on wearable devices

This paper proposed a novel electrocardiogram (ECG) automatic diagnose system for health assistance and rescue related with cardiovascular diseases. This system consists of three parts: 1) Data acquisition subsystem, this subsystem acquires ECG data from wearable devices on users' body and transmit them to the cloud server. 2) Deep learning analysis subsystem, with the help of convolutional neural network, the important feature lied inside ECG signal can be extract for abnormal heart condition detection. Hierarchical residual modules provide the network the ability to see seconds of signal and make a decision through the combination of features. Meanwhile, the global max pooling layer on top of the network enables it to capture the most important feature across the whole ECG signal with periodicity. This subsystem is a crucial part for cardiac status based health caring. 3) Back-stage management subsystem, methodical data storage and management were conducted in this subsystem, which also provides the users an interface to access their healthy data and body status. Assembling these three parts of system, real-time ECG diagnose for people in need and timely medical rescue can be implemented.