SURFACE AND LOCOMOTOR TASK DO NOT AFFECT THE FRACTAL CHARACTERISTICS OF GAIT VARIABILITY

Author ORCID Identifier

0000-0001-8000-6469

Advisor Information

Dr. Aaron Likens

Location

MBSC Ballroom - Poster #508 - G

Presentation Type

Poster

Start Date

4-3-2022 12:30 PM

End Date

4-3-2022 1:45 PM

Abstract

INTRODUCTION

Human walking presents fluctuations from one step to the next, adjusting to the demands of the environment. This gait variability tends to fluctuate over many time scales (milliseconds, seconds, minutes, hours, etc.), due to the interactions among the individual (pathology, stage of learning, etc.), task (easy or difficult), and environment (smooth, rugged, etc.) [1]. Importantly, the temporal structure of gait variability is characterized by fractal patterns in healthy and younger individuals but tend to degrade in older adults and those with neurodegenerative diseases [1-2]. Furthermore, such fractal patterns have been observed not only in walking but also in running gait, while traversing either overground or a treadmill. However, there is no systematic study that has compared both locomotor tasks on both surfaces. In addition, previous studies on human gait variability have focused on monofractal (fractal patterns that do not change over time) aspects of gait, which is limited in its ability to characterize the adaptability of gait to an ever-changing environment [2]. Multifractality (fractal patterns that change over time) has been suggested as an alternative framework that better captures variability in human movements [2]. Therefore, we aimed to investigate how monofractal and multifractal characteristics of human gait variability depend on contexts related to locomotive surface and task. We hypothesized that both task and surface would influence the monofractal and multifractal characteristics present in stride intervals during self-paced trials.

METHODS

Eleven subjects (6 female, age = 31.09 ± 9.96 years, mass = 69.28 ± 18.10 kg) were recruited based on an a priori power analysis [3] to participate in a 2×2 repeated measures experiment, comparing the effects of surface and task on the fractal patterns found in gait variability. Participants walked and ran overground and on a treadmill for the following conditions: Treadmill Walking (TW), Treadmill Running (TR), Overground Walking (OW) and Overground Running (OR). Monofractal patterns of stride-to-stride intervals were assessed using detrended fluctuation analysis (DFA) which produces the scaling exponent alpha, α. When α ~ 1.00, this is consistent with fractal patterns and healthy variability [4]. Multifractal DFA was used to derive a summary of the changes in fractal patterns that occurred during each trial, known as the multifractal width, W. When W is large, this signifies more multifractal characteristics, and is found to be larger in healthy subjects compared to diseased individuals like those with Parkinson’s disease [2].

Each participant’s heel strikes from each condition (TW, TR,OW, OR) and for 20 consecutive minutes were analysed. Stride intervals were computed as the time difference of successive heel strikes of the same leg. DFA and MFDFA algorithms were run on each subject’s stride intervals for all trials, giving one α and one W per trial. Separate 2×2 RM ANOVAs were implemented to assess differences in α and differences in W as a function of surface and task.

RESULTS AND DISCUSSION

Due to data loss and equipment failure, only six (4 female, age = 31.17 ± 13.5 years, mass = 55.79 ± 7.18 kg) of the original eleven subjects’ data were included. The two-way interaction between the surface and task condition on α was not significant, F(1,5)=0.0491, p=.8334. Similarly, the main effects of surface, F(1,5)=.7631, p=.4223, and task, F(1,5)=.0428, p=.8443, were not significant. Further, the partial omega squared effect size indicated that the differences in surface, task and interaction explained 13% (small effect), .848% (small effect), and .972% (small effect) of the fractal behavior observed, respectively [5].

The two-way interaction between the surface and task condition on W was not significant, F(1,5)=3.5643, p=.1177. Similarly, the main effects of surface, F(1,5)=.6501, p=.4567, and task, F(1,5)=.4904, p=.5150, were not significant. Further, the partial omega squared effect size indicated that the differences in surface, task and interaction explained 5% (small effect), 6% (small effect), and 23% (medium effect) of the fractal behavior observed, respectively [5].

These preliminary results fail to replicate known effects of surface, where treadmill walking reduces α [6]. Further, the surface and task effects on W are inconclusive. Pending additional data will determine whether our results question the generality of previous findings in the literature concerning effects of both surface and task [4,6].

REFERENCES

  1. Stergiou, 2020. Biomechanics and Gait Analysis
  2. Cavanaugh et al., 2017. J Neurol Phys Ther. 41: 245-251
  3. Kuznetsov & Rhea, 2017. PLOS One. 12: e0174144
  4. Ravi et al.,2020. Front. Physiol. 11(562)
  5. Cohen, 1998. Statistical Power Analysis for the Behavioral Sciences
  6. Terrier & Deriaz, 2011. J. Neuroeng. Rehabilitation. 8(12)

ACKNOWLEDGEMENTS

This research was supported by an UNO GRACA to TW, NIH P20GM109090 to NS and AL, and NSF 2124918 to NS and AL.

This document is currently not available here.

COinS
 
Mar 4th, 12:30 PM Mar 4th, 1:45 PM

SURFACE AND LOCOMOTOR TASK DO NOT AFFECT THE FRACTAL CHARACTERISTICS OF GAIT VARIABILITY

MBSC Ballroom - Poster #508 - G

INTRODUCTION

Human walking presents fluctuations from one step to the next, adjusting to the demands of the environment. This gait variability tends to fluctuate over many time scales (milliseconds, seconds, minutes, hours, etc.), due to the interactions among the individual (pathology, stage of learning, etc.), task (easy or difficult), and environment (smooth, rugged, etc.) [1]. Importantly, the temporal structure of gait variability is characterized by fractal patterns in healthy and younger individuals but tend to degrade in older adults and those with neurodegenerative diseases [1-2]. Furthermore, such fractal patterns have been observed not only in walking but also in running gait, while traversing either overground or a treadmill. However, there is no systematic study that has compared both locomotor tasks on both surfaces. In addition, previous studies on human gait variability have focused on monofractal (fractal patterns that do not change over time) aspects of gait, which is limited in its ability to characterize the adaptability of gait to an ever-changing environment [2]. Multifractality (fractal patterns that change over time) has been suggested as an alternative framework that better captures variability in human movements [2]. Therefore, we aimed to investigate how monofractal and multifractal characteristics of human gait variability depend on contexts related to locomotive surface and task. We hypothesized that both task and surface would influence the monofractal and multifractal characteristics present in stride intervals during self-paced trials.

METHODS

Eleven subjects (6 female, age = 31.09 ± 9.96 years, mass = 69.28 ± 18.10 kg) were recruited based on an a priori power analysis [3] to participate in a 2×2 repeated measures experiment, comparing the effects of surface and task on the fractal patterns found in gait variability. Participants walked and ran overground and on a treadmill for the following conditions: Treadmill Walking (TW), Treadmill Running (TR), Overground Walking (OW) and Overground Running (OR). Monofractal patterns of stride-to-stride intervals were assessed using detrended fluctuation analysis (DFA) which produces the scaling exponent alpha, α. When α ~ 1.00, this is consistent with fractal patterns and healthy variability [4]. Multifractal DFA was used to derive a summary of the changes in fractal patterns that occurred during each trial, known as the multifractal width, W. When W is large, this signifies more multifractal characteristics, and is found to be larger in healthy subjects compared to diseased individuals like those with Parkinson’s disease [2].

Each participant’s heel strikes from each condition (TW, TR,OW, OR) and for 20 consecutive minutes were analysed. Stride intervals were computed as the time difference of successive heel strikes of the same leg. DFA and MFDFA algorithms were run on each subject’s stride intervals for all trials, giving one α and one W per trial. Separate 2×2 RM ANOVAs were implemented to assess differences in α and differences in W as a function of surface and task.

RESULTS AND DISCUSSION

Due to data loss and equipment failure, only six (4 female, age = 31.17 ± 13.5 years, mass = 55.79 ± 7.18 kg) of the original eleven subjects’ data were included. The two-way interaction between the surface and task condition on α was not significant, F(1,5)=0.0491, p=.8334. Similarly, the main effects of surface, F(1,5)=.7631, p=.4223, and task, F(1,5)=.0428, p=.8443, were not significant. Further, the partial omega squared effect size indicated that the differences in surface, task and interaction explained 13% (small effect), .848% (small effect), and .972% (small effect) of the fractal behavior observed, respectively [5].

The two-way interaction between the surface and task condition on W was not significant, F(1,5)=3.5643, p=.1177. Similarly, the main effects of surface, F(1,5)=.6501, p=.4567, and task, F(1,5)=.4904, p=.5150, were not significant. Further, the partial omega squared effect size indicated that the differences in surface, task and interaction explained 5% (small effect), 6% (small effect), and 23% (medium effect) of the fractal behavior observed, respectively [5].

These preliminary results fail to replicate known effects of surface, where treadmill walking reduces α [6]. Further, the surface and task effects on W are inconclusive. Pending additional data will determine whether our results question the generality of previous findings in the literature concerning effects of both surface and task [4,6].

REFERENCES

  1. Stergiou, 2020. Biomechanics and Gait Analysis
  2. Cavanaugh et al., 2017. J Neurol Phys Ther. 41: 245-251
  3. Kuznetsov & Rhea, 2017. PLOS One. 12: e0174144
  4. Ravi et al.,2020. Front. Physiol. 11(562)
  5. Cohen, 1998. Statistical Power Analysis for the Behavioral Sciences
  6. Terrier & Deriaz, 2011. J. Neuroeng. Rehabilitation. 8(12)

ACKNOWLEDGEMENTS

This research was supported by an UNO GRACA to TW, NIH P20GM109090 to NS and AL, and NSF 2124918 to NS and AL.