A novel approach for acoustic estimation of neck fluid volume between men and women.

Obstructive Sleep apnea can be caused by fluid shift from the legs to the neck that narrows the upper airway (UA) and contributes to changes in tracheal sound. Tracheal sound is generated from the turbulent airflow in the pharynx and respiratory airways and it has recently been used to estimate increases in neck fluid volume (NFV). However, tracheal sound is also highly variable among people, especially across the sexes. In this paper, a novel method is proposed to select tracheal sound features towards estimating NFV in men and women separately. To validate this method, it was applied to the tracheal sound data of 28 healthy individuals. Our proposed feature selection algorithm is based on sparse representations and incorporates NFV to maximize the relevance of selected features. This feature selection eliminates the dependence of the previous methods on calibrating the model for every individual. Two models, regression and Kalman filters, are then used to estimate NFV from selected features. Kalman filter obtains the highest performance, estimating NFV with more than 90% accuracy in both men and women. This algorithm can be used to develop non-invasive acoustic technologies to investigate the effects of fluid on UA anatomy in general applications. These results could be used to develop convenient devices to monitor the neck edema and its contribution to sleep apnea severity in fluid retaining patients such as heart or renal failure.