Smart Sensing System for Combined Activity Classification and Estimation of Knee Range of Motion

This paper proposes a novel wearable system and assesses its reliability in monitoring sagittal knee movement, and discriminating between activities of daily living. The system consists of a flexible conductive polymer unit, embedded into a pair of leggings at the level of the knee, interfaced with a customized sensing node for wireless data acquisition. Design constraints included the need for the system to be unobtrusive, low cost, low power, and simple to use. The wearable system was evaluated through a series of trials conducted on healthy participants, tested on two different occasions, while walking, running, and going up and down a set of stairs. The waveforms of the sensor output resemble typical knee kinematics curves. An intraclass correlation coefficient greater than 0.8 was obtained for the output signal of the sensor from which, knee movement is derived for each of the different activities. Time and frequency domain features of the signal were used to discriminate between activities. Results show good discriminative capacity of selected parameters to an accuracy of 93% when employing a random forest analytical approach. These results suggest that the system can be used accurately to monitor both knee movement and activity performed in unconstrained environments, and thus suggesting its potential use to support knee rehabilitation.