E-Textile smart socks for gait analysis: a preliminary validation study

Abstract

In the last years wearable technologies and smart fabrics are arousing a great deal of interest in the healthcare field. The possibility to directly interact with the body, along with the comfort of fabrics, represent great opportunities to develop interesting applications. Smart sensing socks, able to collect kinematic and/or foot pressure signals, are attractive solutions to perform gait analysis also outside clinical laboratories, e.g., in the home environment for remote continuous monitoring of patients. In this study we analyse the performance of Sensoria smart socks (Sensoria Health Inc. Seattle, WA, USA) in detecting the major spatio-temporal gait analysis metrics. We validate the results provided by the system in comparison with those of the IMU-based gait analysis system OPAL Mobility Lab by APDM (APDM Inc, Portland, OR, USA). Three textile pressure sensors are embedded in the Sensoria Sock to monitor plantar pressure, while the hardware unit is connected on the lateral part of the ankle to measure kinematic signals and transmit data to a smartphone. Twelve records were acquired on five healthy subjects (2 women). Each subject wore the sensing Sensoria smart socks and three OPAL Inertial Measurement Units (two on the feet and one in lumbar position), in order to perform simultaneous recording of the walking. The trial consisted of walking 10 metres, turning and returning to the starting point at preferred speed. The following resulting gait metrics were considered from the two systems and compared: Gait Cycle Time (GCT) [s], Stance Phase [%GCT], Cadence [steps/min]. Statistical comparison was performed by means of Bland-Altman test. Fig. 1 shows the results of the statistical analysis. Fig. 1 a presents the principal statistics of the considered metrics for the two systems. The table in Fig. 1 a also reports the numeric results of Bland- Altman test, expressed as the average bias (Sensoria minus OPAL) and its 95% Confidence Interval (CI) (Lower Bound LB – Upper Bound UB). Fig. 1 b, c and d respectively represent the Bland- Altman Plots for GCT, Stance Phase and Cadence. In this preliminary study we investigate the performances of the wearable Sensoria smart socks in evaluating the major temporal gait metrics. Results underline a general agreement in measuring GCT ( bias = − 0.00593 s ) and Cadence ( bias = − 0.0820 steps / min ). The confidence interval of bias includes the zero for both metrics, indicating that the differences randomly occur and are not systematic. The Bland-Altman plot for Stance Phase metric shows a significant difference between the measures provided by the two systems: Sensoria smart socks underestimate the foot stance phase duration with respect to the reference system OPAL Mobility Lab. In this case the average bias is − 7.75 % and its confidence interval does not include zero. The agreement in GCT values and the simultaneous disagreement of stance phase demonstrate discrepancies between the two systems in the detection of intermediate gait events (initial and terminal contact). In conclusion, smart socks performance in the detection of GCT and cadence are satisfactory, however improvements are needed in order to assess more specific gait metrics. In future studies, we aim to test the socks on a larger population and in the assessment of other temporal and spatial gait metrics.