A machine learning approach for automated recognition of movement patterns using basic, kinetic and kinematic gait...

Area of Science:

• Biomechanical Engineering
• Computational Neuroscience
• Gerontology

Background:

• Aging is associated with significant alterations in gait patterns.
• Objective assessment of gait changes is crucial for understanding age-related functional decline and for developing targeted interventions.
• Traditional gait analysis methods can be labor-intensive and subjective.

Purpose of the Study:

• To investigate the efficacy of a machine learning approach, specifically Support Vector Machines (SVM), for the automatic recognition of gait changes associated with aging.
• To evaluate the performance of SVMs using various gait measures including temporal/spatial, kinetic, and kinematic data.
• To identify the minimal set of gait features required for accurate age-based gait classification.

Main Methods:

• Gait data from 12 young and 12 elderly participants were collected using synchronized motion analysis and force platforms during normal walking.
• Twenty-four gait features were extracted from temporal/spatial, kinetic, and kinematic data.
• Support Vector Machine (SVM) models were developed and tested for generalization performance across six different kernel functions.

Main Results:

• The SVM achieved an overall accuracy of 91.7% in distinguishing between young and elderly gait patterns.
• SVM classification performance was robust across various kernel functions.
• Feature selection identified that as few as three gait features, one from each data type, could achieve 100% accuracy in differentiating age groups.

Conclusions:

• Support Vector Machines demonstrate significant potential for accurate and automated gait classification based on age.
• A parsimonious set of gait features can effectively discriminate between young and elderly individuals.
• This approach holds promise for applications in healthcare, sports science, and assistive technology.