Related Experiment Video
Updated: Jun 1, 2026
07:24
A Method for Quantifying Upper Limb Performance in Daily Life Using Accelerometers
Published on: April 21, 2017
Reliability and validity of bilateral ankle accelerometer algorithms for activity recognition and walking speed after
Bruce H Dobkin1, Xiaoyu Xu, Maxim Batalin
1Department of Neurology, School of Medicine, University of California Los Angeles, Los Angeles, CA 90095, USA. bdobkin@mednet.ucla.edu
Stroke
|June 4, 2011
Summary
New wearable sensors accurately measure real-world mobility and activity in stroke survivors. These advanced algorithms provide reliable data for rehabilitation studies, improving patient monitoring and outcomes.
Area of Science:
- Biomedical Engineering
- Rehabilitation Science
- Wearable Technology
Background:
- Current mobility outcome measures for stroke trials are limited, often confined to laboratory settings and short walks.
- There's a need for continuous, real-world monitoring of daily activity and performance in stroke patients.
- Objective data on community-based activity can inform exercise compliance and skill practice in routine care and clinical trials.
Purpose of the Study:
- To develop and validate machine-learning algorithms using triaxial accelerometers for measuring mobility in stroke survivors.
- To assess the reliability and accuracy of these algorithms in recognizing various physical activities and quantifying walking speed in a community setting.
- To compare the activity patterns of stroke survivors with healthy controls.
Main Methods:
- Twelve adults with hemiparetic stroke and six healthy controls wore ankle-mounted triaxial accelerometers.
- Machine-learning algorithms calculated walking speed, which was compared against stopwatch measurements.
- Algorithm reliability in identifying walking, exercise, and cycling was validated against participant activity logs.
Main Results:
- Algorithm-calculated walking speed showed a high correlation with stopwatch-measured speed (r=0.98, P=0.001).
- The algorithms reliably identified various activities including walking, cycling, stair climbing, and leg exercises.
- Stroke survivors exhibited more sedentary behavior and slower walking speeds compared to healthy controls, with distinct gait patterns.
Conclusions:
- Bayesian algorithms utilizing inertial sensors demonstrate high test-retest reliability and validity for activity pattern recognition.
- This technology offers ratio scale data for real-world monitoring of lower extremity activities and walking speed.
- The findings support the use of these algorithms in stroke and rehabilitation studies for objective outcome measurement.
