A wearable system for respiratory signal filtering based on activity: a preliminary validation

Respiratory Rate (RR) can be detected with wearable devices, which can be used to monitor daily life activities continuously. Here we introduce a configuration that allows to filter the respiratory signal based on the activity level with the same sensor system. One healthy subject was recruited (age 55 yo., weight 80 kg, height 1.76 m, sex M). The protocol involved the assessment of RR in four conditions: standing, walking, running slow, and running fast. Data from a gold standard (K5; COSMED, Rome, Italy) and from the novel IMU-based sensor system developed by the research team were collected. The IMU-based sensor system consists of three units placed respectively on the thorax, on the abdomen, and on the lower back. The analysis of the IMU signals is based on Principal Component Analysis (PCA) to identify different movement patterns. Agreement was found between the K5 and the IMUs system. In a static posture, RR can be computed from the first PCA component. During dynamic activities, RR can generally be computed from the second PCA component. The first PCA component during dynamic activities is related to the stride. This finding suggests further research to develop an algorithm for respiratory signal filtering based on activity, thus improving accuracy.