Respiratory rate and pulse rate can be reliably monitored during dynamic activities with wearables

One of the major challenges in the field of wearable devices is to accurately measure physiological parameters during dynamic activities because of motion artifacts. We designed completely wearable Wireless Body Sensor Network (WBSN) for cardio-respiratory monitoring during dynamic activities. The WBSN is composed of three IMUs to detect the respiratory rate (RR), and of a fourth unit to detect the pulse rate (PR). 30 healthy volunteers (17 men, mean age 25.9±6.0 years, mean weight 68.7±9.7 kg, mean height 170.9±9.5 cm) were enrolled in a validation protocol consisting in walking, running, and cycling. The devices composing the WBSN were compared to reference systems, in particular the Cosmed K5 metabolic cart was used as a reference for RR and a Garmin chest strap for PR. The IMU-based system proved to be particularly effective in monitoring RR during cycling, with a Root Mean Square Error (RMSE) of 3.8 breaths per minute (bpm), and during running, with a RMSE of 4.7 bpm. The respiratory signal during walking exhibited a frequency content like the stride, making it difficult to properly filter the desired signal content; in this case the RMSE was of 5.4 bpm. PR showed good agreement with the reference heart rate monitor (RMSE = 13 beats per minute). The WBSN is therefore a very promising technique that can be used to monitor chronic respiratory patients during daily life activities, without losing information when patients perform dynamic activities.