Validation of a Wearable Device and an Algorithm for Respiratory Monitoring During Exercise

This paper investigates the performances of a head-mounted wearable device for the breath-by-breath monitoring of respiratory frequency (f
_R
) during exercise. The device exploits a new algorithm to estimate f
_R
from the breathing-related pressure drops (ΔP) recorded at the nostrils level. Performances of the wearable device in measuring the breath-by-breath and 30-s average f
_R
values were evaluated during two high-intensity cycling exercise tests performed in the laboratory. ΔP signals were collected from ten volunteers with the wearable device, and the simultaneous measurements with a reference instrument were performed for validation purposes. In addition, numerical simulations were carried out to reproduce the conditions expected in applied settings. Bland-Altman analysis, linear regression (r
²
), and percentage error (%E) were used for comparing the two instruments. Experimental tests demonstrate the robustness and validity of the proposed wearable device and the related algorithm to measure the breath-by-breath f
_R
(overall %E = 4.03%) and 30-s average f
_R
(overall %E = 2.38%) values. Biases obtained with the breath-by-breath analysis (max. -0.06 ± 6.27 breaths/min) were higher than those obtained in the 30-s window analysis (max. -0.03 ± 1.60 breaths/min). In the simulated conditions, %E increased up to 6.65%. The proposed wearable device is suitable for a wide variety of indoor applications where the f
_R
monitoring during exercise at reduced invasiveness is of great value.