Validity and Reliability of Body-Worn Biomechanical Sensor Platform for Gait Analysis

Purpose: Knee osteoarthritis (OA) is currently evaluated clinically using structural imaging. Meaningful change in biomechanical gait variables has been strongly associated with disease progression in knee OA, which can help to advance the understanding of this disease progression. Measurement of gait biomechanics is often performed in laboratories using optical motion capture systems and floor mounted force plates. However, these systems are expensive, not conveniently portable and require skilled staff to operate. Here we describe the development and testing of an alternative approach, a portable biomechanical sensor platform (BSP). The BSP consists of wireless body-worn inertial monitoring units (IMUs) and foot pressure insole units (ISUs), suitable for the measurement of lower extremity gait biomechanics. Appropriate verification and validation is a prerequisite when considering deployment of the BSP for the measurement of gait biomechanics for use as clinical trial endpoints. Methods: The BSP consists of both IMUs (GaitUp, Switzerland) and foot pressure ISUs with integrated IMUs (Moticon, Germany), Figure 1. The BSP is thus capable of measuring spatio-temporal gait parameters, body-segment kinetics and kinematics.Data from the two sensor systems are communicated wirelessly, via Bluetooth TM, where they are integrated and transferred securely to a central database using a custom designed mobile application. To examine the reliability and accuracy of the BSP, two separate studies were conducted. Study 1 involved fifteen healthy participants (mean age 42.9±14.5, 7 female / 8 male) who performed multiple 2-minute (2mWT) and 6 meter (6MWT) walking tests in a laboratory while wearing the BSP. Each subject completed a study routine involving two separate visits to the lab. The reliability of each gait parameters across walking trials was considered within a session (intra-session reliability) and between visits (inter-session reliability). Study 2 evaluated the concurrent validity and reliability of the knee angle measurements and involved fourteen healthy participants (mean age 58.3±4.2, 7 female / 7 male) who completed a study routine involving two separate visits to the lab. Participants performed multiple 6MWT while data were captured simultaneously from the BSP and an optical motion capture system (CODA, Charnwood Dynamics, UK) with integrated floor mounted force plate (Bertec Corp, USA). The IMU-derived measurements were compared to knee angles recorded from the CODA system during the 6MWT. Results: In study 1. Inter-session test-retest reliability for kinetic parameters (Moticon/Insoles only) ranged from good to excellent (ICC>0.75) for parameters such as mean-Local minimum between peaks, to excellent (ICC>0.9) for parameters such as max and mean total force during stance phase. Inter-session test-retest reliability for spatio-temporal (GaitUp/IMU) ranged from moderate to excellent (ICC>0.5) for parameters such as mean stride velocity and stride length, to excellent (ICC>0.9) for parameters such as mean stance, swing time and cadence. Intra-session test-retest reliability for mean kinetic parameters (Moticon/Insoles only) were excellent (ICC>0.9). Intra-session test-retest reliability for spatio-temporal (GaitUp/IMU) ranged from excellent (ICC>0.9) for parameters such as mean stance, swing time and cadence, to moderate to excellent (ICC>0.5) for parameters such as mean stride velocity and stride length. In study 2, moderate (0.5<ICC<0.75) repeatability was observed for knee kinematics with variability estimates in the 5-8 degree range, measured using the shank and thigh IMUs during the 6MWT, Table 1. In study 2, validity statistics, including root-mean-square-error (RMSE) and mean-absolute-error (MAE) for knee angles measured during initial contact and at maximum during gait cycles. Results ranged from 3.83 to 5.37 degrees for the initial contact angle and from 4.49 to 8.85 degrees for the maximum angle. Conclusions: We have evaluated the application of a portable biomechanical sensor platform suitable for the measurement of gait outside of a laboratory setting. Evaluation of the BSP against optical motion capture has demonstrated its validity and reliability in the measurement of several spatio-temporal, kinetic and knee kinematic gait parameters. The reliability of the GaitUp and Moticon sensor systems was examined both within walking trials and between visits for both 6 meter and 2-minute walking tests. For the Moticon insoles, excellent intra-session and inter-session reliability was observed for several gait parameters. Measures based on force (e.g. mean and max total force during stance phase) and spatio-temporal gait parameters (e.g. gait cycle time and stride length) demonstrates reliability when examined across walking trials both within and between visits. For the kinematic measures from the GaitUp sensors, excellent intra-session and inter-session reliability was observed for the gait parameters for each walking trial with slightly weaker reliability observed for the 6MWTs compared to the 2mWTs.