PCA-based SVM for automatic recognition of gait patterns.

Area of Science:

• Biomechanics
• Machine Learning
• Biomedical Engineering

Background:

• Gait analysis is crucial for assessing mobility and detecting age-related changes.
• Accurate classification of gait patterns can aid in early diagnosis and intervention for mobility impairments.

Purpose of the Study:

• To investigate the efficacy of combining Principal Component Analysis (PCA) with Support Vector Machines (SVM) for classifying gait patterns.
• To compare the performance of the PCA-SVM classifier against other methods, such as artificial neural networks.
• To evaluate the impact of different SVM kernels on classification accuracy.

Main Methods:

• Gait data, specifically vertical ground reaction forces, were collected from 30 young and 30 elderly participants during normal walking using a strain gauge force platform.
• Principal Component Analysis (PCA) was employed to extract key features from the gait data.
• Support Vector Machines (SVM) were developed using the PCA-extracted features for gait classification.

Main Results:

• The PCA-SVM classifier achieved an average accuracy of 90% in distinguishing between young and elderly gait patterns.
• The proposed PCA-SVM method demonstrated superior performance compared to an artificial neural network-based classifier.
• SVM with polynomial and radial basis function kernels outperformed SVM with a linear kernel in this classification task.

Conclusions:

• The combination of PCA and SVM provides an effective method for classifying gait patterns based on kinetic data.
• This technique shows significant potential for future clinical applications in gait analysis and mobility assessment.
• The findings highlight the robustness of PCA-SVM for differentiating age-related gait characteristics.