Neural Network Classification of Cardiac Activity Based on Cardiogram Data for Driver Support System

The article describes the use of neural network technologies for predicting humans heart activity within healthcare wireless systems. Developing algorithm could become a part of integrated system of driver support system embedded on board of the automobile. The trained algorithm is a convolutional neural network with memorizing of previous calculations. The model was trained on the original data set and showed poor classification accuracy on classes with a small number of examples. Expanding the number of examples allowed to increase the recognition accuracy of the two weakest classes by 8 and 19 %.