Investigation of Body Locations for Cardiac and Respiratory Monitoring With Skin-Interfaced Inertial Measurement Unit Sensors

Among all wearable technologies, inertial measurement units (IMUs) may be used for estimating both heart rate (HR) and respiratory rate (RR) from chest wall movements/orientations. However, the influence of sensor placement for cardiorespiratory monitoring has not been investigated yet. At this scope, we collected triaxial accelerometer data from 15 healthy volunteers at different at-rest postures (i.e., sitting, lying, and standing) and during walking. This work aims at: 1) investigating the signals captured in five chest’s measurement sites to determine the most informative axis and sensor position at the different postures and to identify the overall best sensor configuration (called general-purpose configuration) independent of the analyzed subject and posture and 2) assessing the performance of the general-purpose configuration in estimating RR and HR against reference values retrieved from a medical-grade wearable device. Finally, the performance of the general-purpose configuration was compared against a posture-dependent configuration, implying the selection of the best axis/sensor placement per each subject and posture. Based on the results, we found that: 1) the sensor positioned at the mitral valve level is the most promising, while the dorsoventral direction is the most informative; 2) considering both at-rest postures and walking, the proposed method allows reaching mean absolute error (MAE) < 1.5 beats/min for HR and always < 4 breaths/min for RR using the general-purpose configuration; and 3) the posture-dependent configuration improved the RR estimation, especially during walking and standing but does not provide any advantages in HR estimation.