Combining electrocardiogram signal with Accelerometer signals for Human Activity Recognition using Convolution neural network

As the environment getting polluted, people are suffering with different medical problems also people are causes about their health as well. Considering this in the mind, Body sensor based human activity recognition attracting researcher towards this direction. A fusion of electrocardiogram signals and accelerometer signals processed through convolution neural network is proposed in this paper. Accelerometer placed at different location of the human body are fused with the electrocardiogram signals, generated through the ECG sensors placed at the chest of the human body. These fused signal are processed through convolution neural network to automatically detect the features and finally apply softmax for classification of the activities. We choose mHEALTH dataset for the experiment and achieve 98.91% validation accuracy. (paper)