Understanding variation in human movement - The good and the bad in motor variability

Abstract

A stable gait pattern and the ability to maintain upright posture are fundamental motor functions to support human bipedal locomotion. Both gait and balance deteriorate in response to age and pathology. In the context of aging western societies, maintaining secure ambulation is key to ensure the independence and mobility of individuals. Therefore, assessment of the quality of gait and balance performance is an essential aspect in the clinical management of aging as well as diseased populations and serves to make diagnosis, monitor disease progression and evaluate treatment effects. All human movements are inherently variable and the assessment of motor variability is thought to be especially sensitive for the evaluation of movement quality and might furthermore relate to the functioning of the sensory-motor system. However, the application and interpretation of motor variability assessment in the clinical practice is still limited due to uncertainty whether variability during movement production has a functional meaning or is simply an unwanted side product of it. Therefore, the relationship between quality of movement and magnitude of motor variability remains elusive. Furthermore, no comprehensive analysis has been performed to identify motor variability threshold levels to distinguish good from bad movement performance. Finally, it remains unknown if the sensory-motor system actively controls motor variability and consequently which nervous system structures might be involved in this task. In a first step, a systematic literature review and meta-analysis was conducted in order to define variability threshold levels to distinguish good from bad movement performance. Based on the assumption that healthy adult subjects perform walking and balancing at an optimal level of variability, whereas subjects with neurological disease at sub-optimal levels, over 13'000 titles comparing healthy and pathological performance were screened and finally 109 publications were included in the analysis. The comparison of 2775 patients and 3001 healthy subjects revealed that 2.6\% of stride time variability and 265mm$^2$ sway area during balancing are threshold values to distinguish healthy from pathological performance. Thereby, it was possible to present clear indices for the clinical identification of pathological movement behavior based on motor variability assessment. In order to investigate if gait variability is uniformly elevated in patients, in a following step, changes in different gait variability measures were observed in response to Parkinson's disease. This study provides evidence for a distinct control of variability in response to pathology, where loss of temporal variability control is compensated by tighter control of spatial variability. Furthermore, in the context of Parkinson's disease the observed distinct gait variability signature was related to the functioning of specific neurophysiological structures and thereby supports the claim to use motor variability measures to evaluate the function of individual entities of the human sensory-motor system. Observation of footfall kinematics or joint angle variability typically serves to evaluate gait variability. These measurable quantities might or might not be representative of clinically relevant characteristics, for example the dynamic stability of walking, which is more suited to determine the quality of the walking pattern. Therefore, in an experimental study, the association between footfall kinematic variability and dynamic stability of walking was investigated by changing the requirements to stabilize the center of mass during treadmill walking. Dynamic stability was assessed by computing the largest Lyapunov exponent of the medio-lateral center of mass movements. Ten structural equation models were established in order to investigate the association between single, as well as combinations of footfall variability measures and the largest Lyapunov exponent. The results indicate that only the combination of spatial and temporal footfall kinematic measures in the sagittal and frontal plane can predict a fraction (6\%) of dynamic stability behavior. Thus, this study highlights that footfall variability measures are insufficient to represent complex measures of gait quality and that comprehensive gait observation is required for the valid evaluation of walking quality in the clinical context. Finally, in an attempt to observe neurophysiological responses during the active control of motor variability, a setup was developed to allow simultaneous assessment of standing variability and integration of Ia-afferent information (H-reflex) for the control of the task. Then, the response of the sensory-motor system to various sensory and biomechanical perturbations was observed in order to assess the importance of the Ia-afferent reflex loop for the control of standing variability. It was shown that 4\% of the effect of loss of visual information on standing variability can be compensated by increased reliance on Ia-afferent information. However, changes in the temporal structures of standing variability together with the significant effect of a dual-task on the standing variability highlight the general superior role of central over peripheral nervous system structures for the control of motor variability. In conclusion, this thesis provides clear threshold values for the clinical identification of pathological walking and balance function based on measures of motor variability. Furthermore, it was shown how the function of distinct neurophysiological structures can be evaluated by comprehensive gait variability assessment. However, complex functions that describe the quality of the walking are not sufficiently represented by individual gait variability measures. In order to make walking quality measurable, it is necessary to establish more complex biomechanical models to depict the interrelationships between fundamental gait functions and individual gait variability measures. Finally, the thesis highlights the role of central nervous system structures for the control of motor variability. --> A stable gait pattern and the ability to maintain upright posture are fundamental motor functions to support human bipedal locomotion. Both gait and balance deteriorate in response to age and pathology. In the context of aging western societies, maintaining secure ambulation is key to ensure the independence and mobility of individuals. Therefore, assessment of the quality of gait and balance performance is an essential aspect in the clinical management of aging as well as diseased populations and serves to make diagnosis, monitor disease progression and evaluate treatment effects. All human movements are inherently variable and the assessment of motor variability is thought to be especially sensitive for the evaluation of movement quality and might furthermore relate to the functioning of the sensory-motor system. However, the application and interpretation of motor variability assessment in the clinical practice is still limited due to uncertainty whether variability during movement production has a functional meaning or is simply an unwanted side product of it. Therefore, the relationship between quality of movement and magnitude of motor variability remains elusive. Furthermore, no comprehensive analysis has been performed to identify motor variability threshold levels to distinguish good from bad movement performance. Finally, it remains unknown if the sensory-motor system actively controls motor variability and consequently which nervous system structures might be involved in this task.In a first step, a systematic literature review and meta-analysis was conducted in order to define variability threshold levels to distinguish good from bad movement performance. Based on the assumption that healthy adult subjects perform walking and balancing at an optimal level of variability, whereas subjects with neurological disease at sub-optimal levels, over 13'000 titles comparing healthy and pathological performance were screened and finally 109 publications were included in the analysis. The comparison of 2775 patients and 3001 healthy subjects revealed that 2.6% of stride time variability and 265mm$^2$ sway area during balancing are threshold values to distinguish healthy from pathological performance. Thereby, it was possible to present clear indices for the clinical identification of pathological movement behavior based on motor variability assessment. In order to investigate if gait variability is uniformly elevated in patients, in a following step, changes in different gait variability measures were observed in response to Parkinson's disease. This study provides evidence for a distinct control of variability in response to pathology, where loss of temporal variability control is compensated by tighter control of spatial variability. Furthermore, in the context of Parkinson's disease the observed distinct gait variability signature was related to the functioning of specific neurophysiological structures and thereby supports the claim to use motor variability measures to evaluate the function of individual entities of the human sensory-motor system.Observation of footfall kinematics or joint angle variability typically serves to evaluate gait variability. These measurable quantities might or might not be representative of clinically relevant characteristics, for example the dynamic stability of walking, which is more suited to determine the quality of the walking pattern. Therefore, in an experimental study, the association between footfall kinematic variability and dynamic stability of walking was investigated by changing the requirements to stabilize the center of mass during treadmill walking. Dynamic stability was assessed by computing the largest Lyapunov exponent of the medio-lateral center of mass movements. Ten structural equation models were established in order to investigate the association between single, as well as combinations of footfall variability measures and the largest Lyapunov exponent. The results indicate that only the combination of spatial and temporal footfall kinematic measures in the sagittal and frontal plane can predict a fraction (6%) of dynamic stability behavior. Thus, this study highlights that footfall variability measures are insufficient to represent complex measures of gait quality and that comprehensive gait observation is required for the valid evaluation of walking quality in the clinical context.Finally, in an attempt to observe neurophysiological responses during the active control of motor variability, a setup was developed to allow simultaneous assessment of standing variability and integration of Ia-afferent information (H-reflex) for the control of the task. Then, the response of the sensory-motor system to various sensory and biomechanical perturbations was observed in order to assess the importance of the Ia-afferent reflex loop for the control of standing variability. It was shown that 4% of the effect of loss of visual information on standing variability can be compensated by increased reliance on Ia-afferent information. However, changes in the temporal structures of standing variability together with the significant effect of a dual-task on the standing variability highlight the general superior role of central over peripheral nervous system structures for the control of motor variability.In conclusion, this thesis provides clear threshold values for the clinical identification of pathological walking and balance function based on measures of motor variability. Furthermore, it was shown how the function of distinct neurophysiological structures can be evaluated by comprehensive gait variability assessment. However, complex functions that describe the quality of the walking are not sufficiently represented by individual gait variability measures. In order to make walking quality measurable, it is necessary to establish more complex biomechanical models to depict the interrelationships between fundamental gait functions and individual gait variability measures. Finally, the thesis highlights the role of central nervous system structures for the control of motor variability.