Fractal Fluctuations in Human Walking: Comparison Between Auditory and Visually Guided Stepping

Terrier, Philippe

doi:10.1007/s10439-016-1573-y

Cited by 52 publications

(79 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The data used in the present study were collected by the author in a previous study aimed at analyzing the influence of synchronization to external sensory cues on gait variability [24,25]. Thirty-six healthy adults participated in the study:…”

Section: Data Collection and Pre-processingmentioning

confidence: 99%

Gait Recognition via Deep Learning of the Center-of-Pressure Trajectory

Terrier

2020

Applied Sciences

Self Cite

View full text Add to dashboard Cite

Fig. 6. CAM analysis. Twenty-four segments were randomly selected from the training set containing the gait data of six participants (left columns). These segments were used to fine-tune the last layers of the convolutional neural network (CNN) that was pre-trained on the training set of the 30 other participants. This CNN classified 30 segments drawn randomly from the test set with 100% accuracy (right columns). Class activation mapping (CAM) was performed on each sample of the test set. Color coding shows which parts of the signals are prioritized to be used by the CNN to perform classification. Warm color (red, orange): high focus; cold colors (green, blue): low focus. AbstractThe fact that every human has a distinctive walking style has prompted a proposal to use gait recognition as an identification criterion. Using end-to-end learning, I investigated whether the center-of-pressure trajectory is sufficiently unique to identify a person with a high certainty. Thirty-six adults walked on a treadmill equipped with a force platform that recorded the positions of the center of pressure. The raw two-dimensional signals were sliced into segments of two gait cycles . A set of 20,250 segments from 30 subjects was used to configure and train convolutional neural networks (CNNs). The best CNN classified a separate set containing 2,250 segments with 99.9% overall accuracy. A second set of 4,500 segments from the six remaining subjects was then used for transfer learning. Several small subsamples of this set were selected randomly and used for fine tuning. Training with two segments per subject was sufficient to achieve 100% accuracy. The results suggest that every person produces a unique trajectory of underfoot pressures and that CNNs can learn the distinctive features of these trajectories. Using transfer learning, a few strides could be sufficient to learn and identify new gaits.

show abstract

Section: Data Collection and Pre-processingmentioning

confidence: 99%

Gait Recognition via Deep Learning of the Center-of-Pressure Trajectory

Terrier

2020

Applied Sciences

Self Cite

View full text Add to dashboard Cite

show abstract

“…In treadmill experiments, if an additional instruction of gait synchronization is superimposed on the task of walking at the belt speed, a generalized anti-persistent pattern is then observed (Terrier & Dériaz, 2012;Roerdink et al, 2015;Choi et al, 2017). This phenomenon exists both when synchronizing stride intervals to a metronome (auditory cueing), and when aligning step lengths to visual cues projected onto the treadmill belt (visual cueing) (Terrier, 2016).…”

Section: Introductionmentioning

confidence: 99%

“…Foot switches, i.e., pressure sensitive insoles, can be used to detect heel strikes on the ground and can thus collect time series of stride intervals Sejdić et al, 2012;Almurad et al, 2018). Several methods using the continuous measure of the positions of various body parts have also been proposed: 1) high-accuracy GPS (Terrier, Turner & Schutz, 2005); 2) 3-D video analysis (Dingwell & Cusumano, 2010); and 3) an instrumented treadmill that records the center-ofpressure trajectory (Terrier & Dériaz, 2012;Terrier, 2016;Roerdink et al, 2019). These methods require a preliminary discretization of the position signals via minima/maxima detection algorithms Roerdink et al, 2008;Dingwell & Cusumano, 2010).…”

Section: Introductionmentioning

confidence: 99%

Complexity of human walking: the attractor complexity index is sensitive to gait synchronization with visual and auditory cues

Terrier

2019

Preprint

Self Cite

View full text Add to dashboard Cite

Background. During steady walking, gait parameters fluctuate from one stride to another with complex fractal patterns and long-range statistical persistence. When a metronome is used to pace the gait (sensorimotor synchronization), long-range persistence is replaced by stochastic oscillations (anti-persistence). Fractal patterns present in gait fluctuations are most often analyzed using detrended fluctuation analysis (DFA). This method requires the use of a discrete times series, such as intervals between consecutive heel strikes, as an input. Recently, a new nonlinear method, the attractor complexity index (ACI), has been shown to respond to complexity changes like DFA. But in contrast to DFA, ACI can be applied to continuous signals, such as body accelerations. The aim of this study was to further compare DFA and ACI in a treadmill experiment that induced complexity changes through sensorimotor synchronization. Methods. Thirty-six healthy adults walked 30 minutes on an instrumented treadmill under three conditions: no cueing, auditory cueing (metronome walking), and visual cueing (stepping stones). The center-of-pressure trajectory was discretized into time series of gait parameters, after which a complexity index (scaling exponent alpha) was computed via DFA. Continuous pressure position signals were used to compute the ACI. Correlations between ACI and DFA were then analyzed. The predictive ability of DFA and ACI to differentiate between cueing and no-cueing conditions was assessed using regularized logistic regressions and areas under the receiver operating characteristic curves (AUROC). Results. DFA and ACI were both significantly different among the cueing conditions. DFA and ACI were correlated (Pearson’s r = 0.78). Logistic regressions showed that DFA and ACI could differentiate between cueing/no cueing conditions with a high degree of confidence (AUROC = 1.0 and 0.96, respectively). Conclusion. Both DFA and ACI responded similarly to changes in cueing conditions and had comparable predictive power. This support the assumption that ACI could be used instead of DFA to assess the long-range complexity of continuous gait signals.

show abstract

“…In healthy adults, the temporal organization of fluctuations may also change under different conditions: during metronomic walking (i.e., stepping in time with an auditory metronome), stride time fluctuations become anti-persistent, i. e., large fluctuations are likely to be followed by smaller fluctuations, and vice-versa [1,9]. Similarly, stride length and stride speed become anti-persistent when healthy young adults step on visual targets or walk on a treadmill, respectively [3,10]. More recent studies also evidenced that stride time fluctuations can become more persistent when gait is paced by visual or auditory cues [9, 12–14].…”

Section: Introductionmentioning

confidence: 99%