We were surprised in this article by Sonenblum et al.  at the lack of comparison with previously published work. We have previously validated this technique  using a tri-axial accelerometer attached to the wheel of the wheelchair, and processing the data in an almost identical manner to that of Sonenblum et al. except with regard to smoothing the signal. We used change in angle between successive sampling points to indicate movement, and then smoothed the signal by defining a minimum duration of 1 s for movement and non-movement periods.
- wheelchair use
- physical activity