Automatic recognition of gait patterns in human motor disorders using machine learning: A review

Area of Science:

• Biomedical Engineering
• Computer Science
• Neurology

Background:

• Automatic human movement recognition is key for assessing abnormal gait patterns.
• Machine learning excels with complex, multidimensional gait data.

Purpose of the Study:

• Compare machine learning algorithms for gait pattern recognition in motor disorders.
• Identify discriminant features from gait dynamics.
• Highlight methods to enhance gait recognition performance.

Main Methods:

• Conducted an electronic literature search across major databases (Web of Science, IEEE, Scopus).
• Keywords included 'human recognition', 'gait patterns', and 'feature selection methods'.

Main Results:

• Kernel principal component analysis and genetic algorithms effectively reduce feature dimensions, handling nonlinear data.
• Support vector machines (SVMs) demonstrated superior accuracy and generalization capabilities.

Conclusions:

• Combining data reduction, cross-validation, and normalization with SVMs provides an objective tool for clinical status assessment.
• Future work includes real-time applications for powered assistive devices in daily life.