Muscle synergy complexity is associated with altered post-stroke gait dynamics Restricted; Files Only

Abstract

Understanding how post-stroke impairments alter the complex neural and biomechanical processes governing gait (i.e., gait dynamics) could inform rehabilitation personalization. We recently showed that a recurrent neural network (RNN) gait dynamics model revealed individual differences in post-stroke gait dynamics, termed gait signatures, that differ from able-bodied (AB) adults. Muscle coordination constraints, as described by muscle synergies, may underlie individual differences in gait signatures. Here I characterized the relationship between post-stroke muscle synergies and gait signatures. I used an existing dataset of 20 AB adults and 55 adults with post-stroke hemiparesis during 30 seconds of treadmill walking at their self-selected speed. Muscle activity was recorded from 8 muscles bilaterally using electromyography (EMG). Sagittal plane joint kinematics were estimated using optical motion capture. I computed individual-specific gait signatures by training an RNN to predict the time evolution of joint kinematics. To quantify the inter-individual similarity of gait signatures, the Euclidean distance between all pairs of gait signatures and between each participant’s gait signature and the average AB gait signature was computed. To quantify muscle coordination, participant- and leg-specific paretic/left-leg muscle synergies were computed using non-negative matrix factorization. To determine if lower post-stroke synergy complexity is associated with increased deviations from AB gait signatures, I compared the similarity of gait signatures to the AB average signature between groups with two, three, and four synergies (one-way ANOVA with Tukey-Kramer post-hoc tests; α = 0.05). To determine if people with more similar muscle synergy structures have more similar gait signatures, I regressed the similarity of gait signatures against the similarity of synergy structures for all participants. Supporting our first prediction, the 2-, 3-, and 4-synergy groups’ gait signatures differed from AB, with the 3- and 4-synergy groups’ gait signatures being more similar to the average AB signature than those of the 2-synergy group (p < 0.005, Cohen’s d > 1.32). Consistent with the relationship between less-complex synergy control and walking speed, more-severe muscle coordination constraints post-stroke correspond to a reduced ability to walk with AB gait dynamics. Synergy structure similarity was not associated with gait signature similarity (r^2 = 0.01). Synergy similarity, therefore, does not uniquely determine the inter-individual similarity of gait dynamics. Individual differences in biomechanical or sensorimotor impairments may also influence the inter-individual similarity of post-stroke gait dynamics. 

Table of Contents

1 Background....................................................................................................................1

1.1 Introduction................................................................................................................1

1.2 Gait Dynamics.............................................................................................................2

1.3 Muscle Coordination....................................................................................................4

1.4 Muscle Synergies May Explain Individual Differences in Gait Dynamics...........................5

2 Methods 6 2.1 Dataset Description...................................................................................6

2.2 Gait Signatures............................................................................................................9

2.3 Muscle Synergies........................................................................................................11

2.4 Comparisons..............................................................................................................13

3 Results.........................................................................................................................14

3.1 Gait signatures, using a larger cohort of stroke survivors, support previous findings........14

3.2 Muscle synergies were similar to previously-reported values.........................................14

3.3 Comparisons..............................................................................................................15

4 Discussion....................................................................................................................18

4.1 Methodological Considerations...................................................................................19

4.2 Implications..............................................................................................................20

4.3 Future Work..............................................................................................................20

Bibliography...................................................................................................................22 

About this Honors Thesis

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Emory College
Department	Physics
Degree	B.S.
Submission	Honors Thesis
Language	English
Research Field	Biophysics, General Health Sciences, Rehabilitation and Therapy Biology, Neuroscience
关键词	Artificial Intelligence and Machine Learning Computational Modeling Neural Dynamics Stroke Rehabilitation
Committee Chair / Thesis Advisor	Gordon J. Berman, Emory University
Committee Members	Lena H. Ting, Emory University Trisha M. Kesar, Emory University

最新修改

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	File download under embargo until 24 May 2026	2024-04-17 12:13:01 -0400	File download under embargo until 24 May 2026

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions