A Novel Approach for Swallow Detection by Fusing Throat Acceleration and PPG Signal

Swallow, as one of the most complex somatic reflexes, is a mirror of our daily ingestion and physical health. In recent years, accelerometry has become a simple and popular tool for non-invasive swallow detection. However, very few physiological signals are used in this field. In this paper, we first notice that swallowing causes significant throat photoplethysmogram (PPG) waveform fluctuation. With this inspiration, we present a novel approach for swallow detection by fusing throat acceleration (ACC) and PPG signal. The support vector machine (SVM) classifier is employed and a score level fusion method is used to access and fuse the information from ACC and PPG signal. The fusion result achieves 90.5% precision as well as 60.0% specificity, higher than the PPG based detection and the ACC based detection. The experiment result shows that our method combines the characteristics of ACC and PPG signal, providing better overall performance in swallow detection.