Assessment of walker-assisted gait based on Principal Component Analysis and wireless inertial sensors

Abstract

Introduction: This study investigates a gait research protocol to assess the impact of a walker model with forearm supports on the kinematic parameters of the lower limb during locomotion. Methods: Thirteen healthy participants without any history of gait dysfunction were enrolled in the experimental procedure. Spatiotemporal and kinematic gait parameters were calculated by using wireless inertial sensors and analyzed with Principal Component Analysis (PCA). The PCA method was selected to achieve dimension reduction and evaluate the main effects in gait performance during walker-assisted gait. Additionally, the interaction among the variables included in each Principal Component (PCs) derived from PCA is exposed to expand the understanding of the main differences between walker-assisted and unassisted gait conditions. Results: The results of the statistical analysis identifi ed four PCs that retained 65% of the data variability. These components were associated with spatiotemporal information, knee joint, hip joint and ankle joint motion, respectively. Conclusion: Assisted gait by a walker model with forearm supports was characterized by slower gait, shorter steps, larger double support phase and lower body vertical acceleration when compared with normal, unassisted walking.

Review on anomalous gait behavior detection using machine learning algorithms

“…PCs with eigenvalues greater than one, 60% to 70% cumulative percentage and variation of data based on scree plot analysis. Thus, four PCs were selected from 31 features of gait variables [23].…”