Classification of daily-life postural transitions using trunk-worn wearable barometric pressure sensor

Distinguishing sedentary from dynamic behavior is essential in addressing disease conditions that are influenced by mobility. Event-based activity recognition algorithms essentially rely on accurate classification of siting and standing postural transitions to distinguish whether the subject is sitting or standing. In this paper, the use of barometric pressure, to estimate altitude, is investigated. It enabled a correct classification of postural transitions with a sensitivity of 92.31% and specificity of 98.06%. The type (sit-to-stand or stand-to-sit) of transition was also accurately identified.