Implementing Gait Kinematic Trajectory Forecasting Models on an Embedded System.

Smart algorithms for gait kinematic motion prediction in wearable assistive devices including prostheses, bionics, and exoskeletons can ensure safer and more effective device functionality. Although embedded systems can support the use of smart algorithms, there are important limitations associated with computational load. This poses a tangible barrier for models with increased complexity that demand substantial computational resources for superior performance. Forecasting through Recurrent Topology (FReT) represents a computationally lightweight time-series data forecasting algorithm with the ability to update and adapt to the input data structure that can predict complex dynamics. Here, we deployed FReT on an embedded system and evaluated its accuracy, computational time, and precision to forecast gait kinematics from lower-limb motion sensor data from fifteen subjects. FReT was compared to pretrained hyperparameter-optimized NNET and deep-NNET (D-NNET) model architectures, both with static model weight parameters and iteratively updated model weight parameters to enable adaptability to evolving data structures. We found that FReT was not only more accurate than all the network models, reducing the normalized root-mean-square error by almost half on average, but that it also provided the best balance between accuracy, computational time, and precision when considering the combination of these performance variables. The proposed FReT framework on an embedded system, with its improved performance, represents an important step towards the development of new sensor-aided technologies for assistive ambulatory devices.

A wearable device-enabled therapeutic approach to improve joint attention in autism spectrum disorder: a prospective pilot study.

It has been previously proposed that interventions aimed at integrating and co-activating music processing and motor control systems could have therapeutic potential for priming social skill development in children with autism spectrum disorder (ASD). In this study, we assessed this hypothesis through a wearable sensor platform called Ambulosono ("Ambulo"-walk; "sono"-sound) in which pleasurable children's musical stimuli are contingently linked to effortful motor action (locomotor step size), thus creating a motivational state proposed to be conducive to joint attention (JA) operation. Five participants were recruited from a community-based partner and were assessed by scoring responses following therapist-directed bids for JA. Multiple assessment sessions through a repeated time-series design were conducted to determine baseline and post-intervention scores. The intervention session consisted of approximately 15 min of Ambulosono exposure. Baseline and post-intervention data were aggregated and analyzed using a linear mixed-effect model. The wearable sensor and wireless headphones of the Ambulosono system were tolerated by the participants, and there were no adverse effects associated with the use of the device. We found an increase in the average responses to bids for JA during the Ambulosono intervention phase compared to baseline across participants. This increase did not appear to result from enhanced general arousal. Our pilot data support feasibility and further testing of Ambulosono as a therapeutic aid for integration into community-based ASD programs to augment shared child-therapist social interactions.

Muscarinic acetylcholine receptor-dependent persistent activity of layer 5 intrinsic-bursting and regular-spiking neurons in primary auditory cortex.

Cholinergic signaling coupled to sensory-driven neuronal depolarization is essential for modulating lasting changes in deep-layer neural excitability and experience-dependent plasticity in the primary auditory cortex. However, the underlying cellular mechanism(s) associated with coincident cholinergic receptor activation and neuronal depolarization of deep-layer cortical neurons remains unknown. Using in vitro whole cell patch-clamp recordings targeted to neurons (n = 151) in isolated brain slices containing the primary auditory cortex (AI), we investigated the effects of cholinergic receptor activation and neuronal depolarization on the electrophysiological properties of AI layer 5 intrinsic-bursting and regular-spiking neurons. Bath application of carbachol (5 µM; cholinergic receptor agonist) paired with suprathreshold intracellular depolarization led to persistent activity in these neurons. Persistent activity may involve similar cellular mechanisms and be generated intrinsically in both intrinsic-bursting and regular-spiking neurons given that it 1) persisted under the blockade of ionotropic glutamatergic (kynurenic acid, 2 mM) and GABAergic receptors (picrotoxin, 100 µM), 2) was fully blocked by both atropine (10 µM; nonselective muscarinic antagonist) and flufenamic acid [100 µM; nonspecific Ca2+-sensitive cationic channel (CAN) blocker], and 3) was sensitive to the voltage-gated Ca2+ channel blocker nifedipine (50 µM) and Ca2+-free artificial cerebrospinal fluid. Together, our results support a model through which coincident activation of AI layer 5 neuron muscarinic receptors and suprathreshold activation can lead to sustained changes in layer 5 excitability, providing new insight into the possible role of a calcium-CAN-dependent cholinergic mechanism of AI cortical plasticity. These findings also indicate that distinct streams of auditory processing in layer 5 intrinsic-bursting and regular-spiking neurons may run in parallel during learning-induced auditory plasticity.NEW & NOTEWORTHY Cholinergic signaling coupled to sensory-driven neuronal depolarization is essential for modulating lasting changes in experience-dependent plasticity in the primary auditory cortex. Cholinergic activation together with cellular depolarization can lead to persistent activity in both intrinsic-bursting and regular-spiking layer 5 pyramidal neurons. A similar mechanism involving muscarinic acetylcholine receptor, voltage-gated Ca2+ channel, and possible Ca2+-sensitive nonspecific cationic channel activation provides new insight into our understanding of the cellular mechanisms that govern learning-induced auditory cortical and subcortical plasticity.

A novel single-sensor-based method for the detection of gait-cycle breakdown and freezing of gait in Parkinson's disease.

Objective measurement of walking speed and gait deficits are an important clinical tool in chronic illness management. We previously reported in Parkinson's disease that different types of gait tests can now be implemented and administered in the clinic or at home using Ambulosono smartphone-sensor technology, whereby movement sensing protocols can be standardized under voice instruction. However, a common challenge that remains for such wearable sensor systems is how meaningful data can be extracted from seemingly "noisy" raw sensor data, and do so with a high level of accuracy and efficiency. Here, we describe a novel pattern recognition algorithm for the automated detection of gait-cycle breakdown and freezing episodes. Ambulosono-gait-cycle-breakdown-and-freezing-detection (Free-D) integrates a nonlinear m-dimensional phase-space data extraction method with machine learning and Monte Carlo analysis for model building and pattern generalization. We first trained Free-D using a small number of data samples obtained from thirty participants during freezing of gait tests. We then tested the accuracy of Free-D via Monte Carlo cross-validation. We found Free-D to be remarkably effective at detecting gait-cycle breakdown, with mode error rates of 0% and mean error rates < 5%. We also demonstrate the utility of Free-D by applying it to continuous holdout traces not used for either training or testing, and found it was able to identify gait-cycle breakdown and freezing events of varying duration. These results suggest that advanced artificial intelligence and automation tools can be developed to enhance the quality, efficiency, and the expansion of wearable sensor data processing capabilities to meet market and industry demand.

Development and Validation of Ambulosono: A Wearable Sensor for Bio-Feedback Rehabilitation Training.

Wearable technology-based measurement systems hold potential for the therapeutic and rehabilitation management of patients with various chronic diseases. The purpose of this study was to assess the accuracy and test⁻retest reliability of a new-generation wearable sensor-based system, dubbed Ambulosono, for bio-feedback training. The Ambulosono sensor system was cross-validated by comparing its functionality with the iPod touch (4th generation) sensor system. Fifteen participants underwent a gait test to measure various gait parameters while wearing both the iPod-based and Ambulosono sensors simultaneously. The physically measured values (i.e., the true values) of step length, distance traveled, velocity, and cadence were then compared to those obtained via the two-sensor systems using the same calculation algorithms. While the mean percentage error was <10% for all measured parameters, and the intra-class correlation coefficient revealed a high level of agreement between trials for both sensor systems, it was found that the Ambulosono sensor system outperformed the iPod-based system in some respects. The Ambulosono sensor system possessed both reliability and accuracy in obtaining gait parameter measurements, which suggests it can serve as an economical alternative to the iPod-based system that is currently used in various clinical rehabilitation programs.

Motivational wheel running reverses cueing behavioural inflexibility in rodents.

Behavioural inflexibility and associated atypical learning behaviours are common clinical manifestations of the autism spectrum disorder (ASD) phenotype. Despite advances in our understanding of ASD, little research has been devoted to experimental interventions that might help to circumvent behavioural inflexibility in ASD. The current paper suggests that motivational locomotion in the form of wheel running can reduce behavioural inflexibility and learning impairments in an ASD rat model, and discusses how the strategy of reward-coupled locomotor activity may lead to clinical interventions for children with ASD.

Somato-dendritic decoupling as a novel mechanism for protracted cortical maturation.

BACKGROUND: Both human and animal data indicate that disruption of the endogenously slow maturation of temporal association cortical (TeA) networks is associated with abnormal higher order cognitive development. However, the neuronal mechanisms underlying the endogenous maturation delay of the TeA are poorly understood. RESULTS: Here we report a novel form of developmental plasticity that is present in the TeA. It was found that deep layer TeA neurons, but not hippocampal or primary visual neurons, exist in a protracted 'embryonic-like' state through a mechanism involving reduced somato-dendritic communication and a non-excitable somatic membrane. This mechanism of neural inactivity is present in intact tissue and shows a remarkable transition into an active somato-dendritically coupled state. The quantity of decoupled cells diminishes in a protracted and age-dependent manner, continuing into adolescence. CONCLUSIONS: Based on our data, we propose a model of neural plasticity through which protracted compartmentalization and decoupling in somato-dendritic signalling plays a key role in controlling how excitable neurons are incorporated into recurrent cortical networks independent of neurogenesis.

Correlation between midline gait function performance and verbal fluency in patients with Parkinson's disease.

BACKGROUND: Verbal fluency impairments are among the earliest and most common cognitive deficits in Parkinson's disease (PD), yet more than one study has shown that as a group, verbal fluency performance is not always different between patients with PD and age-matched controls. Here we examined whether PD-related deficits in speech and language capabilities are detectable in patients who exhibit poor midline motor function, such as control of gait. We postulated that if phonemic verbal fluency (PVF) performance in PD is dissociable from other factors such as age in this patient subgroup, a low PVF will likely reflect a general cognitive-motor deficit attributable to Parkinsonian pathology. METHODS: Thirty-one PD and thirteen controls were evaluated on PVF and the Metronome-Paced Square Step Test (MPSST), a complex sensorimotor task that challenges axial and midline function. The MPSST requires a patient to initiate and maintain a consecutive series of diagonal and midline cross-over steps during the test. RESULTS: We found that despite similar PVF performance between groups, the vast majority of controls but not patients completed the MPSST without any errors, and the odds of completing the MPSST without any errors was 15 times greater among control. More importantly, a significant amount of PVF variability was explained by MPSST performance in PD but not controls even after controlling for age and disease duration. CONCLUSION: Together, these results provide evidence that PVF performance in PD likely reflects a PD-specific process and suggests that the MPSST may be a more sensitive test of PD-specific pathology than PVF.

A new quantitative method for evaluating freezing of gait and dual-attention task deficits in Parkinson's disease.

People with Parkinson's disease (PD) can exhibit disabling gait symptoms such as freezing of gait especially when distracted by a secondary task. Quantitative measurement method of this type of cognitive-motor abnormality, however, remains poorly developed. Here we examined whether stepping-in-place (SIP) with a concurrent mental task (e.g., subtraction) can be used as a simple method for evaluating cognitive-motor deficits in PD. We used a 4th generation iPod Touch sensor system to capture hip flexion data and obtain step height (SH) measurements (z axis). The accuracy of the method was compared to and validated by kinematic video analysis software. We found a general trend of reduced SH for PD subjects relative to controls under all conditions. However, the SH of PD freezers was significantly worse than PD non-freezers and controls during concurrent serial 7 subtraction and SIP tasking. During serial 7 subtraction, SH was significantly associated with whether or not a PD patient was a self-reported freezer even when controlling for disease severity. Given that this SIP-based dual-task paradigm is not limited by space requirements and can be quantified using a mobile tracking device that delivers specifically designed auditory task instructions, the method reported here may be used to standardize clinical assessment of cognitive-motor deficits under a variety of dual-task conditions in PD.

Concurrent arm swing-stepping (CASS) can reveal gait start hesitation in Parkinson's patients with low self-efficacy and fear of falling.

BACKGROUND: Movement incoordination, freezing of gait, fear of falling, low self-efficacy, and multi-tasking can all contribute to falls in Parkinson's disease. How these multi-factorial risks interact in individual patients remain poorly understood. METHODS: Concurrent arm swing-stepping is a simple motor test in which subjects are first asked to swing their arms before being instructed to initiate the secondary task of leg stepping-in-place. We postulated that in patients with multiple fall risks, sensorimotor impairments in upper- and lower-limb movement control can render concurrent arm swing-stepping a demanding dual task, thereby triggering gait hesitation. A total of 31 subjects with Parkinson's disease were enrolled in the study. RESULTS: It was found that concurrent arm swing-stepping induced hesitation primarily in Parkinson's disease patients with low fall-related self-efficacy and a fear of falling. By contrast, concurrent arm swing-stepping led to limb incoordination in both patients and in healthy elderly controls. The calculated specificity and sensitivity of the concurrent arm swing-stepping test was 100 and 42 % for hesitation and 12 and 77 % for incoordination. CONCLUSION: These results suggest that the concurrent arm swing-stepping test can be used in conjunction with conventional psychometric assessments to facilitate multi-factorial assessment of potential fall risk.

Hypothalamic CRH neurons represent physiological memory of positive and negative experience.

Recalling a salient experience provokes specific behaviors and changes in the physiology or internal state. Relatively little is known about how physiological memories are encoded. We examined the neural substrates of physiological memory by probing CRHPVN neurons of mice, which control the endocrine response to stress. Here we show these cells exhibit contextual memory following exposure to a stimulus with negative or positive valence. Specifically, a negative stimulus invokes a two-factor learning rule that favors an increase in the activity of weak cells during recall. In contrast, the contextual memory of positive valence relies on a one-factor rule to decrease activity of CRHPVN neurons. Finally, the aversive memory in CRHPVN neurons outlasts the behavioral response. These observations provide information about how specific physiological memories of aversive and appetitive experience are represented and demonstrate that behavioral readouts may not accurately reflect physiological changes invoked by the memory of salient experiences.

Walking exercise alters pedunculopontine nucleus connectivity in Parkinson's disease in a dose-dependent manner.

Background: Gait disturbances are critical motor symptoms in Parkinson's disease (PD). The mechanisms of gait impairment in PD are not entirely understood but likely involve changes in the Pedunculopontine Nucleus (PPN), a critical locomotion center, and its associated connections. Exercise is universally accepted as helpful in PD, but the extent and intensity of exercise required for plastic changes are unclear. Methods: Twenty-seven PD subjects participated in a 3-month gait training intervention. Clinical assessments and resting-state functional magnetic resonance imaging were performed at baseline and 3 months after exercise. Functional connectivity of PPN was assessed by combining the methods of partial least squares, conditional dependence and partial correlation. In addition, paired t-tests were used to examine the effect of exercise on PPN functional connectivity and clinical measures, and Pearson's correlation was used to assess the association between altered PPN functional connectivity and clinical measures. Results: Exercise significantly improved Unified Parkinson's Disease Rating Scale-III (UPDRS-III). A significant increase in right PPN functional connectivity was observed after exercise, which did not correlate with motor improvement. However, the decrease in left PPN functional connectivity significantly correlated with the improvement in UPDRS-III and was linearly related to both number of walks and the duration of walks. In addition, exercise induced a significant increase in the laterality of PPN connectivity strength, which correlated with motor improvement. Conclusion: PPN functional connectivity is modifiable by walking exercise in both a dose-independent (right PPN and laterality of PPN connectivity strength) and dose-dependent (left PPN) manner. The PPN may contribute to pathological and compensatory processes in PD gait control. The observed gait improvement by walking exercise is most likely due to the reversal of the maladaptive compensatory mechanism. Altered PPN functional connectivity can be a marker for exercise-induced motor improvement in PD.

A versatile computational algorithm for time-series data analysis and machine-learning models.

Here we introduce Local Topological Recurrence Analysis (LoTRA), a simple computational approach for analyzing time-series data. Its versatility is elucidated using simulated data, Parkinsonian gait, and in vivo brain dynamics. We also show that this algorithm can be used to build a remarkably simple machine-learning model capable of outperforming deep-learning models in detecting Parkinson's disease from a single digital handwriting test.

Altering the trajectory of early postnatal cortical development can lead to structural and behavioural features of autism.

BACKGROUND: Autism is a behaviourally defined neurodevelopmental disorder with unknown etiology. Recent studies in autistic children consistently point to neuropathological and functional abnormalities in the temporal association cortex (TeA) and its associated structures. It has been proposed that the trajectory of postnatal development in these regions may undergo accelerated maturational alterations that predominantly affect sensory recognition and social interaction. Indeed, the temporal association regions that are important for sensory recognition and social interaction are one of the last regions to mature suggesting a potential vulnerability to early maturation. However, direct evaluation of the emerging hypothesis that an altered time course of early postnatal development can lead to an ASD phenotype remains lacking. RESULTS: We used electrophysiological, histological, and behavioural techniques to investigate if the known neuronal maturational promoter valproate, similar to that in culture systems, can influence the normal developmental trajectory of TeA in vivo. Brain sections obtained from postnatal rat pups treated with VPA in vivo revealed that almost 40% of cortical cells in TeA prematurely exhibited adult-like intrinsic electrophysiological properties and that this was often associated with gross cortical hypertrophy and a reduced predisposition for social play behaviour. CONCLUSIONS: The co-manifestation of these functional, structural and behavioural features suggests that alteration of the developmental time course in certain high-order cortical networks may play an important role in the neurophysiological basis of autism.

The early overgrowth theory of autism spectrum disorder: Insight into convergent mechanisms from valproic acid exposure and translational models.

The development of new approaches for the clinical management of autism spectrum disorder (ASD) can only be realized through a better understanding of the neurobiological changes associated with ASD. One strategy for gaining deeper insight into the neurobiological mechanisms associated with ASD is to identify converging pathogenic processes associated with human idiopathic clinicopathology that are conserved in translational models of ASD. In this chapter, we first present the early overgrowth theory of ASD. Second, we introduce valproic acid (VPA), one of the most robust and well-known environmental risk factors associated with ASD, and we summarize the rapidly growing body of animal research literature using VPA as an ASD translational model. Lastly, we will detail the mechanisms of action of VPA and its impact on functional neural systems, as well as discuss future research directions that could have a lasting impact on the field.

Visual Gait Lab: A user-friendly approach to gait analysis.

BACKGROUND: Gait analysis forms a critical part of many lab workflows, ranging from those interested in preclinical neurological models to others who use locomotion as part of a standard battery of tests. Unfortunately, while paw detection can be semi-automated, it becomes generally a time-consuming process with error corrections. Improvement in paw tracking would aid in better gait analysis performance and experience. NEW METHOD: Here we show the use of Visual Gait Lab (VGL), a high-level software with an intuitive, easy to use interface, that is built on DeepLabCut™. VGL is optimized to generate gait metrics and allows for quick manual error corrections. VGL comes with a single executable, streamlining setup on Windows systems. We demonstrate the use of VGL to analyze gait. RESULTS: Training and evaluation of VGL were conducted using 200 frames (80/20 train-test split) of video from mice walking on a treadmill. The trained network was then used to visually track paw placements to compute gait metrics. These are processed and presented on the screen where the user can rapidly identify and correct errors. COMPARISON WITH EXISTING METHODS: Gait analysis remains cumbersome, even with commercial software due to paw detection errors. DeepLabCut™ is an alternative that can improve visual tracking but is not optimized for gait analysis functionality. CONCLUSIONS: VGL allows for gait analysis to be performed in a rapid, unbiased manner, with a set-up that can be easily implemented and executed by those without a background in computer programming.

The Effects of Music-Contingent Gait Training on Cognition and Mood in Parkinson Disease: A Feasibility Study.

Background. In Parkinson disease (PD), gait impairments often coexist with nonmotor symptoms such as anxiety and depression. Biofeedback training may improve gait function in PD, but its effect on nonmotor symptoms remains unclear. This study explored the cognitive and global effects of Ambulosono, a cognitive gait training method utilizing step size to contingently control the real-time play of motivational music. Objective. This study examined the feasibility of music-contingent gait training and its effects on neuropsychological test performance and mood in persons with PD. Methods. A total of 30 participants with mild to moderate PD were semirandomized via sequential alternating assignment into an experimental training group or control music group. The training group received 12 weeks of music-contingent training, whereby music play was dependent on the user achieving a set stride length, adjusted online based on individual performance. The control group received hybrid training beginning with 6 weeks of noncontingent music walking, whereby music played continuously regardless of step size, followed by 6 weeks of music-contingent training. Global cognition, memory, executive function, attention, and working memory assessments were completed by blinded assessors at baseline, 6 weeks, and 12 weeks. Motor function, mood, and anxiety were assessed. Results. Average training adherence was 97%, with no falls occurring during training sessions. Improvements on cognitive measures were not clinically significant; however, significant decreases in depression and anxiety were observed in both groups over time (P < .05). Conclusions. Music-contingent gait training is feasible and safe in individuals with PD. Further investigation into potential therapeutic applications of this technology is recommended.

Wearable technological platform for multidomain diagnostic and exercise interventions in Parkinson's disease.

Physical activity and exercise have become a central component of medical management of chronic illness, particular for the elderly who suffer from neurodegenerative disorders that impair their cognition and mobility. This chapter summarizes our recent research showing that a new generation of wearable technology can be adopted as diagnostic and rehabilitation tools for people living with Parkinson's disease. For example, wearable device-enabled 6-min walking test can be automated to eliminate human supervision and many other technical factors that confound the results with conventional testing. With reduced cost and increased test standardization, the technology can be adopted for population-based screening of cardiovascular fitness and gait rehabilitation training efficacy associated with many medical conditions. The Ambulosono platform for multidomain exercise intervention, in particular, has the potential to deliver lasting clinical benefits in slowing PD progression. The platform, through the integration of brisk walking with behavioral shaping strategies such as contingency reinforcement, anticipatory motor control and musical motivational stimulation, creates a home exercise regime that can transform monotonous walking into a pleasurable daily activity and habit.

Differentiating cognitive or motor dimensions associated with the perception of fall-related self-efficacy in Parkinson's disease.

In Parkinson's disease (PD), concurrent declines in cognitive and motor domain function can severely limit an individual's ability to conduct daily tasks. Current diagnostic methods, however, lack precision in differentiating domain-specific contributions of cognitive or motor impairments based on a patients' clinical manifestation. Fear of falling (FOF) is a common clinical manifestation among the elderly, in which both cognitive and motor impairments can lead to significant barriers to a patients' physical and social activities. The present study evaluated whether a set of analytical and machine-learning approaches could be used to help delineate boundary conditions and separate cognitive and motor contributions to a patient's own perception of self-efficacy and FOF. Cognitive and motor clinical scores, in conjunction with FOF, were collected from 57 Parkinson's patients during a multi-center rehabilitation intervention trial. Statistical methodology was used to extract a subset of uncorrelated cognitive and motor components associated with cognitive and motor predictors, which were then used to independently identify and visualize cognitive and motor dimensions associated with FOF. We found that a central cognitive process, extracted from tests of executive, attentional, and visuoperceptive function, was a unique and significant independent cognitive predictor of FOF in PD. In addition, we provide evidence that the approaches described here may be used to computationally discern specific types of FOF based on separable cognitive or motor models. Our results are consistent with a contemporary model that the deterioration of a central cognitive mechanism that modulates self-efficacy also plays a critical role in FOF in PD.

Mechanisms of Hierarchical Cortical Maturation.

Cortical information processing is structurally and functionally organized into hierarchical pathways, with primary sensory cortical regions providing modality specific information and associative cortical regions playing a more integrative role. Historically, there has been debate as to whether primary cortical regions mature earlier than associative cortical regions, or whether both primary and associative cortical regions mature simultaneously. Identifying whether primary and associative cortical regions mature hierarchically or simultaneously will not only deepen our understanding of the mechanisms that regulate brain maturation, but it will also provide fundamental insight into aspects of adolescent behavior, learning, neurodevelopmental disorders and computational models of neural processing. This mini-review article summarizes the current evidence supporting the sequential and hierarchical nature of cortical maturation, and then proposes a new cellular model underlying this process. Finally, unresolved issues associated with hierarchical cortical maturation are also addressed.