Online Learning of Gait Models From Older Adult Data.

This paper proposes a novel approach for online, individualized gait analysis, based on an adaptive periodic model of any gait signal. The proposed method learns a model of the gait cycle during online measurement, using a continuous representation that can adapt to inter- and intra-personal variability by creating an individualized model. Once the algorithm has converged to the input signal, key gait events can be identified based on the estimated gait phase and amplitude. The approach is implemented and tested on retirement home resident 6 min walk (6MW) data using wearable accelerometers at the ankle. The proposed approach converges within approximately four gait cycles and achieves 3% error in detecting initial swing events.11 An early version of this work was presented in [1]. A more extensive description of related work and an extended method, including optimization of learning rates, were added to this paper. Further, this paper applies and evaluates the method to a new and much larger gait dataset taken from older adults who each have a variety of medical conditions. Therefore, the experimental protocol was also updated and the results are entirely novel.

Cortical control of anticipatory postural adjustments prior to stepping.

Human bipedal balance control is achieved either reactively or predictively by a distributed network of neural areas within the central nervous system with a potential role for cerebral cortex. While the role of the cortex in reactive balance has been widely explored, only few studies have addressed the cortical activations related to predictive balance control. The present study investigated the cortical activations related to the preparation and execution of anticipatory postural adjustment (APA) that precede a step. This study also examined whether the preparatory cortical activations related to a specific movement is dependent on the context of control (postural component vs. focal component). Ground reaction forces and electroencephalographic (EEG) data were recorded from 14 healthy adults while they performed lateral weight shift and lateral stepping with and without initially preloading their weight to the stance leg. EEG analysis revealed that there were distinct movement-related potentials (MRPs) with concurrent event-related desynchronization (ERD) of mu and beta rhythms prior to the onset of APA and also to the onset of foot-off during lateral stepping in the fronto-central cortical areas. Also, the MRPs and ERD prior to the onset of APA and onset of lateral weight shift were not significantly different suggesting the comparable cortical activations for the generation of postural and focal movements. The present study reveals the occurrence of cortical activation prior to the execution of an APA that precedes a step. Importantly, this cortical activity appears independent of the context of the movement.

Do measures of reactive balance control predict falls in people with stroke returning to the community?

OBJECTIVE: To determine if reactive balance control measures predict falls after discharge from stroke rehabilitation. DESIGN: Prospective cohort study. SETTING: Rehabilitation hospital and community. PARTICIPANTS: Independently ambulatory individuals with stroke who were discharged home after inpatient rehabilitation (n=95). MAIN OUTCOME MEASURES: Balance and gait measures were obtained from a clinical assessment at discharge from inpatient stroke rehabilitation. Measures of reactive balance control were obtained: (1) during quiet standing; (2) when walking; and (3) in response to large postural perturbations. Participants reported falls and activity levels up to 6 months post-discharge. Logistic and Poisson regressions were used to identify measures of reactive balance control that were related to falls post-discharge. RESULTS: Decreased paretic limb contribution to standing balance control [rate ratio 0.8, 95% confidence interval (CI) 0.7 to 1.0; P=0.011], reduced between-limb synchronisation of quiet standing balance control (rate ratio 0.9, 95% CI 0.8 to 0.9; P<0.0001), increased step length variability (rate ratio 1.4, 95% CI 1.2 to 1.7; P=0.0011) and inability to step with the blocked limb (rate ratio 1.2, 95% CI 1.0 to 1.3; P=0.013) were significantly associated with increased fall rates when controlling for age, stroke severity, functional balance and daily walking activity. CONCLUSIONS: Impaired reactive balance control in standing and walking predicted increased risk of falls post-discharge from stroke rehabilitation. Specifically, measures that revealed the capacity of both limbs to respond to instability were related to increased risk of falls. These results suggest that post-stroke rehabilitation strategies for falls prevention should train responses to instability, and focus on remediating dyscontrol in the more-affected limb.

Predictors of low bone mineral density of the stroke-affected hip among ambulatory individuals with chronic stroke.

UNLABELLED: Risk of hip fracture is greater poststroke than in an age-matched healthy population, in part because of declining hip BMD. We found that individuals may be at risk of loss of hip BMD from muscle atrophy, asymmetrical gait, and poor affected-side ankle dorsiflexor strength. These impairments may be targeted during rehabilitation. INTRODUCTION: This study aimed to determine predictors of low hip BMD on the stroke-affected side in people living in the community. METHODS: Forty-three participants (female; 27.9%), mean age 62.4 ± 13.5 and 17.9 ± 32.8 months, poststroke with motor impairments underwent dual energy X-ray absorptiometry scans. Gait characteristics, isometric strength, body composition, and fasting plasma lipids were measured. RESULTS: At entry, 34.9% (15/43) of the participants had low total hip BMD on the stroke-affected side. Of those with low BMD, 93.3% (14/15) had a step length symmetry ratio >1, indicating greater reliance on the non-paretic leg for weight bearing. Logistic regression analysis revealed that lower affected-side ankle dorsiflexor strength (ß = 0.700, p = 0.02), lower total body fat-free mass index (ß = 0.437, p = 0.02), and greater step length symmetry ratio during walking (ß = 1.135 × 10(3), p = 0.03) were predictors of low hip BMD. CONCLUSION: Low BMD of the stroke-affected side hip is prevalent in over a third of individuals with lower limb motor impairments. These individuals may be at particular risk of accelerated loss of BMD at the hip from asymmetrical gait pattern and poor affected-side ankle dorsiflexor strength. These impairments are intervention targets that may be addressed during rehabilitation which includes resistance training and addresses gait impairments.

Time to disengage: holding an object influences the execution of rapid compensatory reach-to-grasp reactions for recovery from whole-body instability.

Rapid reach-to-grasp (RTG) reactions are important for balance recovery. Despite the benefit of having hands free to regain balance, people do not always release a handheld object. We investigated whether reluctance to release is related to central nervous system (CNS) processing delays that occur when the initial reaction is to drop the object rather than RTG. Young adults sat in a custom-designed chair that tilted backwards. Participants regained balance by reaching to a handle with hands free or while holding onto (1) a chair-fixed object or (2) a SMALL or LARGE free-moving object (unbreakable plastic tubes). EMG was collected from the upper limb to determine onset of reaction. Kinematic data from a digitized wrist marker were used to determine movement time. 9 of 10 participants released the object in every trial. Extensor digitorum onset occurred significantly later than anterior deltoid onset in all conditions. LARGE object release induced further delays in extensor onset while both SMALL and LARGE object release increased response and movement time. Object disengagement led to delays in perturbation-evoked, RTG reactions, particularly in the focal muscle (extensor digitorum) and when the objects' properties posed greater risk for a failed RTG response. We propose that time required for cognitive disengagement accounts for the observed delays. This study offers a potential explanation for the tendency to avoid disengaging from a handheld object during balance recovery. Results also provide insight into the challenges imposed upon the CNS during temporally urgent movements.

Perturbation-evoked cortical activity reflects both the context and consequence of postural instability.

The cerebral cortex may play a role in the control of compensatory balance reactions by optimizing these responses to suit the task conditions and/or to stimulus (i.e. perturbation) characteristics. These possible contributions appear to be reflected by pre-perturbation and post-perturbation cortical activity. While studies have explored the characteristics and possible meaning of these different events (pre- vs. post-) there is little insight into the possible association between them. The purpose of this study was to explore whether pre- and post-perturbation cortical events are associated or whether they reflect different control processes linked to the control of balance. Twelve participants were presented temporally-predictable postural perturbations under four test conditions. The Block/Random tasks were designed to assess modifiability in CNS gain prior to instability, while the Unconstrained/Constrained tasks assessed responsiveness to the magnitude of instability. Perturbations were evoked by releasing a cable which held the participant in a forward lean position. The magnitude of pre-perturbation cortical activity scaled to perturbation amplitude when the magnitude of the perturbation was predictable [F(3,11)=2.906, P<0.05]. The amplitude of pre-perturbation cortical activity was large when the size of the forthcoming perturbation was unknown (13.8 + or - 7.9, 11.4 + or - 9.9, 16.9 + or - 9.3, and 16.1 + or - 10.6 muV for the Block Unconstrained and Constrained and Random Unconstrained and Constrained, respectively). In addition, N1 amplitude scaled to perturbation amplitude regardless of whether the size of the forthcoming perturbation was known (30.1 + or - 17.7, 11.4 + or - 7.1, 30.9 + or - 18.4, 12.4 + or - 6.1 muV). This is the first work to examine modifiability in the pre-perturbation cortical activity related to postural set alterations. The cerebral cortex differentially processes independent components prior to and following postural instability to generate compensatory responses linked to the conditions under which instability is experienced.

Challenging the brain: Exploring the link between effort and cortical activation.

To better understand the contributions of effort on cortical activation associated with motor tasks, healthy participants with varying capacities for isolating the control of individual finger movements performed tasks consisting of a single concurrent abduction of all digits (Easy) and paired finger abduction with digits 2 and 3 abducted together concurrently with digits 4 and 5 (Hard). Brain activity was inferred from measurement using functional magnetic resonance imaging. Effort was measured physiologically using electrodermal responses (EDR) and subjectively using the Borg scale. On average, the Borg score for the Hard task was significantly higher (p=0.007) than for the Easy task (2.9+/-1.1 vs. 1.4+/-0.7, respectively). Similarly, the average normalized peak-to-peak amplitude of the EDR was significantly higher (p=0.002) for the Hard task than for the Easy task (20.4+/-6.5% vs. 12.1+/-4.9%, respectively). The Hard task produced increases in sensorimotor network activation, including supplementary motor area, premotor, sensorimotor and parietal cortices, cerebellum and thalamus. When the imaging data were subdivided based on Borg score, there was an increase in activation and involvement of additional areas, including extrastriate and prefrontal cortices. Subdividing the data based on EDR amplitude produced greater effects including activation of the premotor and parietal cortices. These results show that the effort required for task performance influences the interpretation of fMRI data. This work establishes understanding and methodology for advancing future studies of the link between effort and motor control, and may be clinically relevant to sensorimotor recovery from neurologic injury.

Perturbation-evoked electrodermal activity responds to instability, not just motor or sensory drives.

OBJECTIVE: To determine whether electrodermal responses (EDRs) evoked by postural perturbations were sensitive to the context of compensatory balance control, or simply reflected sensory or motor components of the reaction. METHODS: Thirteen participants were perturbed backwards in an upright chair and (1) performed compensatory reach-to-grasp movements to a handhold to recover balance (COMP); (2) received the perturbation only and the chair stopped via mechanical support (SENS); and (3) performed rapid self-initiated reach-to-grasp movements without perturbation (MOT). RESULTS: EDRs were most frequent and largest in the COMP task, observed in 100% of trials (1.42+/-0.16 microS), compared to 39% of SENS trials (0.31+/-0.12 microS, p<0.0001) and 85% of MOT trials (0.98+/-0.25 microS, p=0.073). EDRs in the MOT task followed two patterns across individuals, leading to post-hoc division of subjects into groups (smaller EDRs than COMP task, n=7, versus equivalent EDRs to COMP task, n=6). Motor patterns were equivalent in both groups, indicating that EDRs did not co vary with efferent drive. CONCLUSIONS: Perturbation-evoked EDRs are not a direct reflection of sensory input or motor drive. SIGNIFICANCE: These findings suggest that evoked autonomic activity may play a functional role in compensatory postural control.

Cortical responses associated with the preparation and reaction to full-body perturbations to upright stability.

OBJECTIVE: To examine the cortical activity associated with 'central set' preparations for induced whole-body instability. METHODS: Self-initiated and temporally unpredictable perturbations to standing balance were caused by the release of a load coupled to a cable affixed to a harness while participants stood on a force plate. Electroencephalographic and electromyographic signals were recorded. RESULTS: Peak activity was located at the Cz electrode. The predictable condition elicited a DC shift 950 ms prior to perturbation onset and was 18.0+/-10.5 micro V in magnitude. Pre-perturbation activity was not associated with the motor act of perturbation initiation and was dissociable from cortical activity related to anticipatory postural muscle activation. Following perturbation onset, N1 potentials were observed with a peak amplitude of 17.6+/-7.2 micro V and peak latency of 140.1+/-25.9 ms. In unpredictable trials, pre-perturbation activity was absent. The peak amplitude (32.0+/-14.8 micro V) and latency (156.5+/-11.8 ms) of the post-perturbation N1 potential were significantly larger (p=0.002) and later (p<0.001) than for predictable trials. CONCLUSIONS: Self-initiated postural instability evokes cortical activity prior to and following perturbation onset. Pre-perturbation cortical activity is associated with changing central set to modulate appropriate perturbation-evoked balance responses. SIGNIFICANCE: These findings establish a link between reactive balance control and cortical activity that precedes and follows perturbations to stability.

Cognitive demands and cortical control of human balance-recovery reactions.

A traditional view has been that balance control occurs at a very automatic level, primarily involving the spinal cord and brainstem; however, there is growing evidence that the cerebral cortex and cognitive processing are involved in controlling specific aspects of balance. The purpose of this review is to summarize recent literature pertaining to the cognitive demands and cortical control of balance-recovery reactions, focussing on five emerging sources of evidence: 1) dual-task studies demonstrating that concurrent performance of cognitive and balance-recovery tasks leads to interference effects; 2) dual-task studies that have examined the temporal dynamics associated with the reallocation of cognitive resources to the balance-recovery task; 3) visual attention studies that have inferred contributions of visual attention based on gaze measurements and/or manipulations to occlude vision; 4) measurements of brain potentials evoked by postural perturbation; and 5) use of transcranial magnetic stimulation to alter contributions from specific cortical areas.

Changes in gait variability during different challenges to mobility in patients with traumatic brain injury.

Postural stability may be compromised in patients who have sustained a traumatic brain injury (TBI). The purpose of the present study was to examine dynamic stability during gait by measuring spatial and temporal variability of foot placement, and to determine the effect of increased difficulty of the walking task on gait variability in patients with TBI. It was hypothesized that patients with TBI will show increased variability in step time, step length, and step width in comparison to healthy controls and that such differences would be accentuated by increased task difficulty. Participants (patients: n=20, controls: n=20) were asked to walk across a pressure sensitive mat at their preferred pace (PW), as fast as possible (FW), and with their eyes closed (EC). In accordance with the hypotheses, patients had significantly greater variability in step time and step length in comparison to healthy controls, and when the complexity of the gait task increased (FW and EC tasks). Although step width variability showed no significant difference between the groups, both control and patient groups had increased step width variability in the EC task. It is proposed that such increases in variability reflect greater challenges to maintaining dynamic stability during gait among individuals with TBI and when performing more difficult tasks.

Is there a trade-off between cognitive and motor recovery after traumatic brain injury due to competition for limited neural resources?

The status of neurorehabilitation for traumatic brain injury (TBI) is under active debate because of a dearth of research findings demonstrating effectiveness. This may be due, in part, to limitations in our understanding of basic mechanisms of cognitive and motor recovery,including those that might impede recovery. In this regard, we examined whether overall recovery following TBI might be undermined by competition between cognitive and motor functions for finite neural resources during recovery. In this preliminary study, 21 moderately and severely impaired patients were administered cognitive and motor assessments at 1, 4, and 12 months post-TBI, and recovery of cognitive and motor functions was measured using regression residuals. Negative correlations between recovery of cognitive versus motor functions were used as the index of competition. We found suggestive evidence that there may indeed be a trade-off between the recovery of cognitive and motor functions after TBI. Implications for rehabilitation are discussed.

The effect of directional compatibility on the response latencies of ocular and manual movements.

Visuomotor coordination is essential for the successful performance of everyday activities, and it could be affected by the directional compatibility between ocular and manual movements. Many tasks, such as driving or operating devices in the workplace, require a variety of coordination patterns with different levels of compatibility between the eyes and the hand. For example, the movement of the eyes and the arm can be coupled when both effectors point towards the same direction whereas in other tasks the movement of the eyes and the arm can be dissociated, for instance when a peripheral object is foveated while a button press response is executed concurrently. The objective of this study was to examine the latency of ocular and manual movements in tasks characterized by variations in directional compatibility. Four tasks were used to manipulate compatibility: 1. point and look at a peripheral stimulus (POINT AND LOOK)--high directional compatibility; 2. point to a peripheral stimulus while fixating in the center (POINT AND FIXATE)--low directional compatibility; 3. press a button while looking at a peripheral stimulus (PRESS AND LOOK)--low directional compatibility; and 4. press a button while fixating in the center (PRESS AND FIXATE)--no directional motor requirement. We hypothesized that the latency of (1) manual and (2) ocular responses would be faster in the task with high directional compatibility compared with the tasks with low compatibility or the task with no directional motor component. Ten healthy participants executed pointing and pressing movements with and without concurrent eye movement to randomly presented visual stimuli. In agreement with the first hypothesis, results showed that in a task with high directional compatibility, manual responses were initiated significantly faster when compared with the tasks with low compatibility or a task with no directional motor component: 1. pointing while looking was initiated 22 ms faster on average than pointing while fixating; 2. pointing while looking was initiated 91 ms faster than pressing accompanied by an eye movement; and 3. pointing while looking was initiated 102 ms faster than pressing while fixating. The second hypothesis was partially supported by data which showed that eye movements directed toward peripheral stimuli were initiated significantly more slowly (30 ms on average) when accompanied by pressing in comparison with the latency of eye movements in the high-compatibility task. In contrast with the hypothesis, eye movements that were accompanied by pointing were not initiated faster than those in a task which required looking without pointing. In summary, these data suggest that directional compatibility is an important aspect of motor control. The effects of directional compatibility are discussed in a conceptual framework that considers the neurophysiological substrates that might be involved in mediating these effects.

Cortical activation following a balance disturbance.

Although recent work suggests that cortical processing can be involved in the control of balance responses, the central mechanisms involved in these reactions remain unclear. We presently investigated the characteristics of scalp-recorded perturbation-evoked responses (PERs) following a balance disturbance. Eight young adults stabilized an inverted pendulum using their ankle musculature while seated. When perturbations were applied to the pendulum, subjects were instructed to return (active condition) or not return (passive condition) the pendulum to its original stable position. Primary measures included peak latency and amplitude of early PERs (the first negative peak between 100 and 150 ms, N1), amplitude of late PERs (between 200 and 400 ms) and onset and initial amplitude of ankle muscle responses. Based on the timing of PERs, we hypothesized that N1 would represent sensory processing of the balance disturbance and that late PERs would be linked to the sensorimotor processing of balance corrections. Our results revealed that N1 was maximal over frontal-central electrode sites (FCz and Cz). Average N1 measures at FCz, Cz, and CPz were comparable between active and passive tasks ( p>0.05). In contrast, the amplitude of late PERs at Cz was less positive for the active condition than for the passive ( p<0.05). The similarity in N1 between tasks suggests a sensory representation of early PERs. Differences in late PERs may represent sensorimotor processing related to the execution of balance responses.

Tactile stimulus predictability modulates activity in a tactile-motor cortical network.

Manipulating objects in the hand requires the continuous transformation of sensory input into appropriate motor behaviour. Using a novel vibrotactile device combined with fMRI, the cortical network associated with tactile sensorimotor transformations was investigated. Continuous tactile stimuli were delivered in a random or predictable pattern to the second digit on the right hand of all subjects. To better distinguish sensory and motor processes, subjects were instructed to make proportionate motor gripping responses with their left hand. A consistent cortical network of activation was revealed that included the supplementary motor, dorsal and ventral premotor, posterior parietal, primary and secondary somatosensory and primary motor cortex. Tracking the unpredictable versus predictable tactile stimulus led to greater delays in motor responses and to increased performance errors. Cortical effects due to stimulus predictability were observed in several components of the network, though it was most evident as increased cortical activation in frontal motor regions during tracking of unpredictable tactile stimuli. In contrast to the proposed hypotheses, primary and secondary somatosensory cortices contralateral to tactile input did not reveal enhanced responses during unpredictable tracking. Facilitation during unpredictable tracking was also observed in primary somatosensory cortex contralateral to motor responses, the receptive site for movement-related afference. The present study provides a novel and controlled approach to investigate the loci associated with tactile-motor processing and to measure the task-specific effect of stimulus predictability on network components.

Adaptation in the motor cortex following cervical spinal cord injury.

BACKGROUND: The nature of the adaptive changes that occur in the cerebral cortex following injury to the cervical spinal cord are largely unknown. OBJECTIVE: To investigate these adaptive changes by examining the relationship between the motor cortical representation of the paretic right upper extremity compared with that of the tongue. The tongue was selected because the spinal cord injury (SCI) does not affect its movement and the cortical representation of the tongue is adjacent to that of the paretic upper extremity. METHODS: FMRI was used to map cortical representations associated with simple motor tasks of the right upper extremity and tongue in 14 control subjects and 9 patients with remote (>5.5 months) cervical SCI. RESULTS: The mean value for the site of maximum cortical activation during upper limb movement was identical between the two groups. The site of maximum left hemispheric cortical activation during tongue movement was 12.8 mm (p < 0.01) medial and superior to that of control subjects, indicating the presence of a shift in cortical activation. CONCLUSION: The findings indicate that the adult motor cortex does indeed adapt following cervical SCI. The nature of the adaptation and the underlying biological mechanisms responsible for this change require further investigation.

Cognitive demands of executing postural reactions: does aging impede attention switching?

A new dual-task paradigm was used to investigate age-related differences in attentional dynamics during rapid balancing reactions evoked by small, unpredictable antero-posterior platform movements. The perturbations were delivered while subjects performed a continuous visuo-motor pursuit-tracking task. Onset of significant deviation in tracking was inferred to indicate switching of attentional resources between tracking and balancing tasks. Although tracking deviation was equally likely to occur subsequent to postural perturbation in healthy young and older adults, deviation onset was delayed, on average, by 67% (123 ms) in the older subjects. Delay in onset of tracking deviation correlated with subsequent delay in generating the peak stabilizing postural response at the ankle. These results suggest that impaired attentional dynamics may exacerbate postural instability in older adults.

Medio-lateral balance adjustments preceding reflexive limb withdrawal are modified by postural demands.

We have recently observed medio-lateral balance adjustments (BA) preceding reflexive stepping elicited by noxious stimulation. While task specific modulation is evident for BA prior to voluntary leg movement, it is unclear whether rapid BA reactions (prior to 'reflexive' stepping) represent a generic response to evoked limb withdrawal or can be modified to suit task-conditions. This study was designed to establish whether the CNS is able to modify rapid onset latency BAs to match task conditions. Reflexive stepping was evoked by applying a noxious stimulus (50 ms stimulus train, 1 ms pulses, 300 Hz, 4 x perceptual threshold) to the plantar surface of the either the left or right foot. Task conditions were varied prior to stimulation by having subjects maintain one of three different static positions: (1) lean left (70% body weight (BW) on left), (2) neutral (50% BW both sides), (3) lean right (70% BW on right). BAs were denoted by centre-of-pressure (CoP) excursions towards the swing foot after the onset of noxious stimulation (average onset latency of 128 ms). There was a significant increase in frequency of occurrence and a significant increase in magnitude of CoP shift when the stimulation was applied to a loaded limb (leaning with 70% BW on the stimulated foot) as compared to an unloaded limb (30% BW). In addition, 78% of loaded trials featured steps taken with the unstimulated foot, which delayed removal of the stimulated foot. Collectively, the results indicate modifiability of the very rapid onset balance adjustments that precede the onset of limb withdrawal revealing complex control of balance exists even over very brief latencies.

Passive and active lower-limb movements delay upper-limb balance reactions.

This study investigated the influence of rhythmic lower-limb activity on the timing of upper-limb balance reactions. Compensatory grasping reactions were evoked in healthy subjects by rapid sagittal tilts of a chair under three conditions: (1) active leg pedaling, (2) passive (motor-driven) leg pedaling, and (3) no lower-limb movement (control task). Compared with control trials, both active and passive pedaling resulted in similar delays in the initiation (43-47 ms) and execution (12-17 ms) of grasping reactions. The similarity between effects due to active and passive movement suggests that the conditioning arose predominantly from sensory discharge associated with lower-limb movement. These results may have important implications for understanding the influence of locomotion or other ongoing movement on the control of stability.

New devices to deliver somatosensory stimuli during functional MRI.

A new class of devices are described for improving investigation of somatosensory neuronal activation using fMRI. Dubbed magnetomechanical vibrotactile devices (MVDs), the principle of operation involves driving wire coils with small oscillatory currents in the large static magnetic field inherent to MRI scanners. The resulting Lorentz forces can be oriented to generate large vibrations that are easily converted to translational motions as large as several centimeters. Representative data demonstrate the flexibility of MVDs to generate different well-controlled vibratory and tactile stimuli to activate special proprioceptive and cutaneous somatosensory afferent pathways. The implications of these data are discussed with respect to the literature on existing devices for producing sensorimotor activation, as well as expanding the scope of current fMRI investigations.