Classification of acoustic physiological signals based on deep learning neural networks with augmented features

Digital signal processing techniques have been applied to analyze physiological signals for decades. Recent progresses in other fields, such as computer vision and machine learning, are attracting people to utilize such technologies for analyzing physiological signals. In this paper, recurrent neural network, often used in deep learning for time series signals, is applied to detect anomalies in heart sound. We successfully detected anomalies with 80% accuracy when augmenting the signals with other features.