A multicentre analysis using a cluster of inertial measurement units for remote monitoring human motion and automated detection of trips and slips

Abstract-Text (inkl. Referenzen und Bildunterschriften)

For remote monitoring and evaluation of human motion during daily life, accurate extraction of kinematic quantities of body segments is of great importance. Current approaches use inertial sensors that require numerical time differentiation to access the angular acceleration vector¹, a mathematical operation that greatly amplifies noise in the acceleration coordinate values. With this multicentre research we had two objectives: i) to introduce a wearable sensor (IMC) that utilises a spatially defined cluster of inertial measurement units for directly measuring the angular acceleration vector and compare the IMC with a numerically derived angular acceleration signal using experimental data; ii) to use the IMC for analysing trunk dynamics and automatic detection of trips and slips. The latter experimental set-up involved standardised perturbations during treadmill walking and less controlled conditions, i.e. perturbed overground walking simulating real-world trips and slips more faithfully as well as various activities of daily living (ADL). The characteristics of the experimental data corresponded to those observed in the simulated sensor data, with the IMC showing a conceptual advantage to determine the angular acceleration vector compared to current inertial sensor solutions. Using a cross-institutional analysis of the various signal sequences via an automated evaluation algorithm including individual thresholds, the results demonstrated that the algorithm was capable to automatically identify all 152 slips and trips revealing 100% sensitivity. Out of 450 ADL tasks the algorithm revealed seven false-positive detections (98% specificity) confirming the feasibility for automatically and remotely detecting trip- and slip-like perturbations. The IMC therefore presents an opportunity to pioneer reliable extraction of kinematic quantities of body segments in daily life and for fall risk assessment in healthy and pathological conditions. ¹Bet et al. 2019, Int J Med Inform