Use of Phase Discontinuity Features in Pocket-Worn Accelerometer to detect Coughs

Cough is considered to be the major symptom of most respiratory problems and due to the recent COVID-19 pandemic cough detection has become a major area of interest. This paper addresses the issue of cough detection using a simple detection methodology using Fast Fourier Transform (FFT) analysis peak detection technique on accelerometer data worn in shirt pocket where smart phones are commonly carried. This methodology measures the peaks on the negative side of the FFT plots indicating phase discontinuity due to coughs or lack of best results are seen on accelerometer signals along directions perpendicular to the chest surface (Z axis) and the vertical height (Y axis) of the test subjects in the different frequency bands 1,2,3 (1-2Hz, 2-4Hz and 4-6Hz). The results indicate that the number peaks in bands 2 and 3 (2-4Hz and 4-6Hz) are significantly higher for detecting cough patterns quite consistently in 2 test subjects of height and weight 1-1.8 meters and 16–64 kilograms.