Human Gait-labeling Uncertainty and a Hybrid Model for Gait Segmentation
Motion capture systems are widely accepted as ground-truth for gait analysis and are used for the validation of other gait analysis systems.To date, their reliability and limitations in manual labeling of gait events have not been studied.
Objectives: Evaluate human manual labeling uncertainty and introduce a new hybrid gait analysis model for long-term monitoring.
Methods: Evaluate and estimate inter-labeler inconsistencies by computing the limits-of-agreement; develop a model based on dynamic time warping and convolutional neural network to identify a valid stride and eliminate non-stride data in walking inertial data collected by a wearable device; Gait events are detected within a valid stride region afterwards; This method makes the subsequent data computation more efficient and robust.
Results: The limits of inter-labeler agreement for key
gait events of heel off, toe off, heel strike, and flat foot are 72 ms, 16 ms, 22 ms, and 80 ms, respectively; The hybrid model's classification accuracy for a stride and a non-stride are 95.16% and 84.48%, respectively; The mean absolute error for detected heel off, toe off, heel strike, and flat foot are 24 ms, 5 ms, 9 ms, and 13 ms, respectively.
Conclusions: The results show the inherent label uncertainty and the limits of human gait labeling of motion capture data; The proposed hybrid-model's performance is comparable to that of human labelers and it is a valid model to reliably detect strides in human gait data.
Significance: This work establishes the foundation for fully automated human gait analysis systems with performances comparable to human-labelers.