Abstract
Continuous monitoring of respiratory patterns and physical activity levels can be useful for remote health management of patients with conditions such as heart disease and chronic obstructive pulmonary disease. In a clinical setting, spirometers serve as the gold standard for monitoring respiratory patterns such as breathing rate and changes in lung volume. However, direct measurements using a spirometer requires placement of a sensor in the patient's airway and is thus infeasible for continuous monitoring in nonclinical, ambulatory settings. Under these conditions, indirect respiration monitoring using electrical impedance plethysmographs (EIP) is more suitable but are susceptible to motion artifacts. In this paper, we investigate whether multichannel EIP can be used to perform virtual spirometry under ambulatory settings. The experiments presented in this paper are based on preliminary data collected from 19 adult human subjects under realistic ambulatory and nonambulatory settings. We first highlight the salient features of the signal acquired from a standard spirometer. We then compare the performance of different biosignal processing algorithms in estimating the spirometer signal using multiple EIP sensors and in the presence of motion artifacts and real-world interferences. We demonstrate that in addition to reliably determining different respiratory patterns and states, multichannel EIP could also be used to reliably extract information regarding different patient physical activity states like bending or stretching.