http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=7523737&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D7523737
Activity counting has demonstrated strong correlations to recovery
before and after stroke rehabilitation. However, there are only
moderate to poor correlations with movement specific features (such as
timing and repetition) that are significant to stroke rehabilitation,
allowing room for improvement. This paper explores the physical meaning
of an accelerometric based activity count, by using a precise tri-axial
accelerometer and tri-axial gyroscope during tasks based on selected
activities of daily living (ADLs). The impact of processing algorithms
and sensor choice were also considered. Nine healthy participants
performed a series of free-world upper extremity movement tasks modelled
after ADLs as well as tasks constrained by speed and direction. Raw
gyroscope and accelerometer data were linearly regressed with medically
graded actigraphy bands for comparison. The results demonstrated that
wrist motion during upper extremity tasks had similar distributions of
data across all planes and axes of motion. The results also highlighted
that processing algorithms based on mean and median epoched data were
more sensitive (p < 0.05) to differences in planes and axes of
motion, but that variance based methods presented lower
root-mean-square-errors (RMSE) errors when linearly regressed with
medically graded technology. The findings from this study help to better
understand inertial patterns of upper extremity rehabilitation based
tasks and physical interpretations of activity count me asures.
Published in:
2016 6th IEEE International Conference on Biomedical Robotics and Biomechatronics (BioRob)Date of Conference:
26-29 June 2016- Page(s):
- 870 - 875
No comments:
Post a Comment