Accelerometry-enabled measurement of walking performance with a robotic exoskeleton: a pilot study.

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Clinical Biomechanics

Background:

• Current clinical scores for exoskeleton walking often rely on single variables, limiting their discriminatory power, especially with small patient cohorts.
• Evaluating walking skills in patients with spinal cord injury (SCI) using lower limb exoskeletons requires robust metrics.
• A novel score is proposed to enhance the assessment of walking ability in SCI patients using powered exoskeletons.

Purpose of the Study:

• To develop and validate a new score that combines multiple features for assessing walking ability in SCI patients using powered exoskeletons.
• To compare the discriminatory power of the new multi-feature score against a standard single-feature outcome measure.

Main Methods:

• Four SCI patients were trained to walk with the ReWalk™ exoskeleton, with body accelerations recorded via wearable accelerometers.
• Four distinct features evaluating walking skills were computed from the acceleration data.
• A Gaussian naïve Bayes surprise score was calculated, comparing patient performance against expert user data.

Main Results:

• All patients demonstrated improvement in walking skills over the training period, with their scores approaching those of expert users.
• The Gaussian naïve Bayes surprise score showed significantly higher discriminative power compared to a score based solely on the number of steps.
• At the end of training, 3 out of 4 patients were statistically differentiated from expert users using the combined score (p < .001).

Conclusions:

• Integrating multiple features into a scoring system offers a more robust metric for evaluating patient progress with robotic exoskeletons.
• The developed multi-feature score demonstrates superior ability to differentiate patient skill levels compared to traditional single-variable measures.
• Future research should explore this approach with additional features and larger patient cohorts.