工口影院

Citation

Byerley J, Mason R, Baker A, Powell D, Pearson L, Barry G, Godfrey A, Mancini M, Stuart S & Morris R (2022) Validation of a low-cost wearable sensor to assess turning in healthy adults. Physiotherapy, 114 (Supplement 1), pp. e112-e113. https://doi.org/10.1016/j.physio.2021.12.062

Abstract
Purpose: Gait characteristics such as turning are commonly impaired in neurological conditions such as Parkinson's disease and traumatic brain injury (TBI). Traditionally, these characteristics have been measured in a laboratory setting using expensive 3D motion capture or research-grade inertial sensor systems, such as the Opal (APDM Inc., Portland, OR, USA). Indeed, algorithms to interpret turning have been previously validated for use in a research-grade sensor (Opal, version 1). Despite validation, high costs associated with research grade devices have prohibited widespread clinical deployment. As such there is demand for validation of low-cost wearable sensors that can be widely deployed in low resource settings. This study aimed to validate turning assessment with a low-cost inertial sensor (Axivity AX6), by simultaneously capturing and comparing to turn algorithm output from the previously validated Opal sensor (Research-grade ‘gold-standard’ reference measure), during several turning tasks in healthy young adults. Methods: Thirty healthy young adults (18 males, 12 females; aged 23.1 ± 5.5 years) wore an AX6, (accelerometer; 100 Hz, ± 16 g, gyroscope; 2000 deg/s, weight 11grams, Axivity, Newcastle-upon-Tyne, UK) and an Opal (accelerometer 128 Hz, ± 6 g, gyroscope; 2000 deg/s, weight 24 g, version 1) sensor on their waist (lumbar L5 region) while they performed 3 mobility assessments; 8 laps of a turns course including turns at 45°, 90° and 135°; a two-minute walk between two lines set 5m apart completing a 180° turn at each line; turning 360° continuously back and forth, on the spot, for 2 min. Turning was assessed using a previously validated custom-made Matlab (MathWorks Inc., Natick, Massachusetts, USA) algorithm, and turn outcomes included number, duration, angle, peak velocity and jerk. All data were analysed using SPSS (version 26, IBM). Intra-class correlation coefficients (ICC) were used to assess the absolute agreement between the turn outcomes from Axivity and Opal sensors. Results: Agreement between the outcomes from the Axivity AX6 and Opal sensors was strongest between the two sensors during the turning 360° continuously, with good to excellent agreement shown for turn duration, angle, peak velocity and jerk (all ICCs > 0.85). There was slightly less agreement for the two-minute walk task, with good agreement for all turn characteristics (all ICCs > 0.80), with the exception of the moderate agreement for turn angle (ICC 0.683). Agreement for turn outcomes was moderate to good during the turns course (ICCs range: 0.58–0.87), with lowest agreement for turn duration. Conclusion(s): This study demonstrated that a low-cost wearable sensor, Axivity AX6, had moderate to excellent agreement with a previously validated research-grade sensor, Opal (version 1), when measuring turns in healthy young adults. Our findings suggest that the low-cost Axivity AX6 sensor is valid tool for assessment of turning outcomes, particularly during continuous turning tasks. Impact: Assessment of turning using low-cost wearable devices could enable adoption of objective digital assessment technology into clinical practice, which could detect mobility impairments in a range of populations. Future research is needed to further assess the validity of the Axivity AX6 sensor in turning assessment in clinical populations, such as Parkinson's disease or mild traumatic brain injury.

Keywords
Wearable; Turning; Validation

Journal
Physiotherapy: Volume 114, Issue Supplement 1