Cough-DL: A Deep Learning Model for Ear-Worn Cough Detection.
Journal:
Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
PMID:
40039966
Abstract
Cough serves as a crucial bio-marker for evaluation and monitoring of pulmonary conditions. With growing interest towards automatic cough detection systems, it's important to acknowledge the existing hurdles on the way for a robust cough counter. These include high false positive rate caused by cough-like sounds in the environment, reduced sensitivity due to background noise interference. In this work, our objective is to tackle these obstacles through a comprehensive exploration of diverse strategies, including signal processing enhancements, innovative data augmentation techniques, and refined modeling approaches with emphasis on specificity to make the model robust in field environment. Our best model achieves sensitivity of 87.29% and specificity of 98.38%, while having a small footprint of 1.6 MB.