Search

In this paper, a neural approach based on using Long-Short Term Memory (LSTM) neural networks is proposed to diagnose patients suffering from PD. In this study, it is shown that the temporal patterns of the gait cycle are different for healthy persons and patients. Therefore, by using a recurrent structure like LSTM, able to analyze the dynamic nature of the gait cycle, the proposed method extracts the temporal patterns to diagnose patients from healthy persons. Utilized data to extract the temporal shapes of the gait cycle are based on changing vertical Ground Reaction Force (vGRF), measured by 16 sensors placed in the soles of shoes worn by each subject. To reduce the number of data dimensions, the sequences of corresponding sensors placed in different feet are combined by subtraction. This method analyzes the temporal pattern of time-series collected from different sensors, without extracting special features representing statistics of different parts of the gait cycle. Finally, by recording and presenting data from 10 seconds of subject walking, the proposed approach can diagnose the patient from healthy persons with an average accuracy of 97.66%, and the total F1 score equal to 97.78%.

Parkinson's Disease (PD) affects a significant amount of elderlies around the world. The progression and effectiveness of medication can be inferred from changes or improvements of gait in patients. Vertical ground reaction force is a common used measure to classify gait. In this study, such data from a publicly available dataset is used to classify gait between PD patients and healthy control subjects of similar ages. The data were preprocessed by normalization, splitting into standard time units and then converted to images. Classification was done using a stacked 2-dimensional and 1-dimenisonal convolutional neural network. Our model achieved 88.7% accuracy, an improvement of 5.3% over the next best performing model that was published.

Gait analysis is crucial in the medical sphere, sports and research. It can facilitate prevention of diseases having gait-related symptoms, alleviation of pain with posture correction and performance improvement. It is being simplified and made more accessible with the development of sensor insoles combined with intelligent analytics. Stridalyzer INSIGHT is a smart insole system which offers ubiquitous clinical-grade gait analysis. In this paper, we present an overview of the device and the embedded sensor network, and evaluation of the data results. Weight distribution, vertical ground reaction force (GRF) and ground contact time (GCT) data has been evaluated using statistical metrics, with the pressure plate as the gold standard. The percentage difference in weight distribution data between the insoles and pressure plate were found to be 7.75(0.78) and −3.85(5.87) for left and right respectively. The correlation between the insoles and pressure plate vertical GRF data for dynamic gait (walking) was found to be 0.65(0.07) and 0.9 for left and right respectively. The percentage difference in GCT data between the insoles and pressure plate were found to be 0 and 12.4(1.8) for left and right respectively. The accuracy of the data can be improved by reducing the capacitance of the sensors and the circuit, compensating for the temporal and magnitudinal effects of capacitance while processing data and increasing the sensor area. Stridalyzer INSIGHT smart insolescan provide out-of-clinic gait analysis to complement the clinical systems, but the data needs to be validated for more varied anthropometric measurements.

Gait analysis has always been an important area of research with its applications on clinical diagnosis, monitoring and rehabilitation. Any pathology to the musculoskeletal or neurological of the lower limb will cause a change in gait and the most common symptom being asymmetrical gait. For this study, interlimb coordination of asymmetrical gait induced by wearing a knee brace during stance phase was assessed through Normalized Symmetry Index (SInorm) of the vertical ground reaction force (VGRF), joint angles and moments of the hip, knee and ankle joint. Asymmetry was noted in the joint angles and moments but not in VGRF during abnormal gait. Of joint angles and moments, joint angles provided a clearer pattern of change in the waveform of the SInorm.

This paper presents an assessment of insole based gait feature variation with progression of Parkinson's disease (PD). Gait and postural disturbances are the common cause of disability and impairment for patients with PD. Variation in gait features can serve as important bio-markers to identify severity of PD. Insole based gait features like stride, swing time, maximum vertical ground reaction force (VGRF) and Centre of Pressure (CoP) variation have been evaluated from PD gait database for a set of PD patients and age matched control group. PD group has been further divided in ‘Mild’ and ‘Moderate’ group-based on severity of PD. Inter group segregation has been derived using three conventional PD progression scales named Hoehn-Yahr, Unified Parkinson's Disease Rating Scale (UPDRS) and Timed Up and Go test (TUG). Gait features were re-evaluated for each group and further analysed to understand the effect and variation of the gait features with proaression of the disease.

This paper presents a novel calibration method of insole sensors to estimate vertical ground reaction force (vGRF) during walking using the Wii Balance Board (WBB) as an accessible alternative to costly force plates. The sensor forces were weighted using linear least-squares regression to fit to the measurements of the WBB during walking and standing. By comparison with force plates, the accuracies of the vGRF estimated by the method with walking through the WBB were equivalent to the force plate-based calibration. In addition, the method using standing on the WBB did not have large deteriorations when compared to the above methods.

Using information about the values of the total center of pressure (CoP) and total vertical ground reaction force (GRF) during gait, quantitative parameters based on CoP trajectory length and trajectory length of 3-D plot of vertical GRF vs. CoP can be calculated and used to compare pathological prosthetic gait and normal gait. Seven patients with right transtibial amputation using passive transtibial prostheses, and ten healthy subjects participated in the study. After calculating the parameters of each patient and healthy subject, the two-sample t-test was used to assess the significance of the differences between the results of the two groups. Also, the observed parameters were compared with each other. For this purpose, Pearson correlation coefficients between the parameters were calculated to study the relationships between the parameters. The results show that in all cases, the parameters described in patients are significantly different than in healthy subjects. New parameters provide new information on way of walking and show the differences in gait step as a whole. The described techniques and calculated and statistically evaluated values of proposed parameters can be used for gait analysis in clinical practice.

Although numerous studies have analyzed the relationship between manual material handling (MMH) and the forces acting on the lumbar spine, the difference in the MMH between experts and novices through the analysis of measured data has not been well studied. The purpose of this paper was to analyze the difference in the MMH positions between ten skilled experts working at a freight transport company (Group 1) and five unskilled novices without any experience (Group 2) during asymmetric lifting. All the human subjects performed asymmetric lifting experiments with closed eyes; the experiments involved moving loads (6 and 18 kg) to the left side. Time series data of the vertical ground reaction force were measured, using a Wii Balance Board, and then, the center-of-pressure (CoP) trajectories were calculated. The balance board was used for the measurement, because it was reliable, inexpensive, and portable and provided good repeatability even on rough surfaces, and all the information pertaining to the load and worker under various conditions was captured without any omissions. Under the 18 kg load condition, the CoP positions for Group 2 were located on the same side during left asymmetric lifting; however, those for Group 1 were located on the opposite side during left asymmetric lifting (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$P, Although numerous studies have analyzed the relationship between lifting of heavy materials and the forces acting on the lumbar spine, differences in lifting strategies between experts and novices have not been well-studied. This paper presents the results of a study that compared the lifting strategies and vertical ground reaction force (vGRF) and center-of-pressure (CoP) trajectories during asymmetric lifting by experts and novices. Unlike novices, most skilled experts position their hips in such a way that their CoP trajectories move toward the opposite side during left asymmetric lifting.

In this paper, we introduce an experiment evaluating performance of football players in countermovement jumps (CMJs). Three methods including time domain, frequency domain, and machine learning algorithms are proposed for performance evaluation. Correlation coefficients and p-values are given for time domain and frequency domain methods, and prediction errors are given for different machine learning algorithms.

In this paper, we propose a novel fatigue assessment technique using Vertical Ground Reaction Force (VGRF) signals acquired during ultra-marathon running. To achieve that goal, we performed frequency-domain on singlechannel VGRF signals on different time intervals during ultra-marathon running and tracked the features' variation. Our method was applied on single-channel VGRF signals recorded via instrumented treadmill during a 24-hour ultramarathon ran by 12 well-trained athletes. The purpose behind our work is to develop computationally-efficient techniques for real-time fatigue assessment during running activities, taking advantage of the low level of complexity and deterministic behavior of the signals in hand when compared to other fatigue indicating electrophysiological signals.

In this paper, we propose guidelines and techniques for optimally handling Vertical Ground Reaction Force (VGRF) signals acquired during running. For that endeavor, we developed an algorithm that performs leg-specific signal separation in an adaptive manner such that each VGRF signal is decomposed into two equally-sized time series each being specific to one of the two legs. Our technique was applied on single-channel VGRF signals recorded via instrumented treadmill during a 24-hour marathon performed by 12 athletes. The purpose of proposing such techniques lies in the fact that the VGRF signal carries combined data relevant to both legs' anatomical and physiological states and its analysis may only describe overall running characteristics, neglecting the importance of inter-leg symmetry in rehabilitation and performance analysis which requires leg-specific analysis.

In this paper, we analyze Vertical Ground Reaction Force (VGRF) signals recorded from normal subjects as well as from subjects attained with Parkinson's disease (PD). The aim of this study is to identify abnormal gait patterns in order to detect patients who are potentially attained with PD. This is done by extracting various significant features from sensors located at 16 different positions on the right and left foot. Finally, extracted features are used to classify between health control and PD subjects and predict the Parkinson phase. Results have shown that extracting parameters based on the summation of sensors output of each foot may hide the conveyed information of VGRF signals. Moreover, frequency-related features are able to classify between PD subjects with different Hoehn and Yahr stages. Finally, results have revealed the importance of power distribution in VGRF signals that vary with PD stages.

Signal is a physical quantity we can measure like gait vertical ground reaction force. However, the latter are such non- stationary signals require a deep understanding of their instantaneous amplitude, phase and frequency. From this, one can model its stationary and non- stationary part and approximate the noise. In addition, one can practice such features for inter-subject classification of the vertical ground reaction force signals like between normal and pathological. Not to add, one objective could also concentrate in intra-subject classification like between usual gait and gait associated with cognitive tasks for the same subject as this paper mainly concerns. For that purpose, Synchrosqueezing of time-frequency representation is being used to spot its power in non-stationary signal analysis and classification. This technique also helped in developing an accurate detection of outliers within such time series signal like when subjects encounter turning points during walking. All this would help in a correct assessing treatment effectiveness and precising the stage of disease. In addition, this would be a starting point for having accurate parameters in elderly fall detection.

A research work is reproducible when all research artifacts such as as text, data, figure and code are available for independent researchers reproduce the results. In this paper, we present a reproducible gait analysis to track Parkinson's Disease evolution by monitoring walking abnormalities. Weapplied Principal Component Analysis into gait data to detect user's abnormalities that may indicate the progression of Parkinson's Disease. We validated our approach with a public database of foot sensor data, which includes vertical ground reaction force records of subjects with healthy gait and Parkinson's Disease patients. We used the euclidean distance asdata classifier. We reached a classification accuracy of 81.00% with leave-one-out cross-validation, which demonstrates the feasibility of our approach for tracking PD's symptoms based on user gait. All relevant data to reproduce our results are available in a public web page.

The development of prosthetic foot components often incorporates mechanical characterization methods that simulate the loading conditions expected in activities of daily living. However, it is recognized that these conditions vary among users and the effect of user variability on mechanical testing results is not currently understood. The objective of this study was to statistically characterize the vertical ground reaction force (vGRF) waveform of prosthesis users to describe its variability in a population by: (1) Applying Principal Component Analysis (PCA) to measured waveforms in a cohort of trans-tibial subjects, and (2) Simulate an expected sample of waveforms for this population using a Monte Carlo method. Phase 1: Three prosthesis users walked on a level walkway at self-selected walking speeds under four prosthetic foot conditions. PCA performed on the vGRF waveforms for affected-limb footsteps resulted in three principal components (PCs) accounting for 91.5% of data variability. Results showed low variability for the same subject using similar designs of prosthetic feet and distinct differences when using a familiar device. Phase 2: Monte Carlo simulation was used to predict a family of 30 vGRF waveforms representative of the sample population. Variability was highest in regions of weight acceptance, mid-stance, and push-off, while lower variability was observed in the transition regions prior to, between, and after these regions. Conclusion: The study supports the use of PCA to describe variability in vGRF waveforms of trans-tibial prosthesis users. The analysis is suitable for Monte Carlo simulation, which showed vGRF waveforms with distinct regions of high and low variability.

One of the effective strategies for designing a muscle-driven humanoid robot that obtains dynamic motion such as jumping is by referring arrangement of muscles in the human body. In order to design a robot that realizes higher performance for vertical jumping, a robot that has hip, knee, and ankle joint equipping not only six monoarticular muscles but also three biarticular muscles is adopted, and the effect of the biarticular muscles on vertical jumping is investigated. In the experiment, robots with/without biarticular muscles are adopted, and an optimized parameter set including the firing period and delay of each muscle is searched using a stochastic searching algorithm. By comparing the performance of the two types of robot models under some sets of conditions, we found that the biarticular muscles relatively contribute to the generation of a vertical ground reaction force for a vertical jumping.

This paper addresses the frequency content of Vertical Ground Reaction force (VGRF) signals. Signal processing of such data is crucial in order to characterize the gait itself. The outcome processing can be used to show the presence of disease, to track the rehabilitation or even to detect the possibility of falling. In order to come up with the correct processing, it is important to set up the rules that cover gait VGRF. In this paper, an attention is given to the frequency content of the VGRF underneath different locations of the foot. The extracted information is then used to compare gait between normal and Parkinson's subjects. Results have shown the existence of three dominant frequencies. Furthermore, it was revealed that the power of frequency is high for the second peak in normal gait compared to Parkinson.

This paper aims to identify gait patterns between normal healthy adults and PD patients based on three vertical ground reaction force gait features namely Maximum vertical heel contact (FZ1), Vertical minimum midstance force (FZ2) and FZ3 known as Maximum vertical push off force. Based on the gait data extracted, it was found that all three vertical GRF features are higher for normal subjects as compared to PD group. Conversely, longer time is taken by PD group to complete a stance phase. Moreover, based on Analysis of Variance, it was also confirmed that all three vertical GRF features are statistical significant. Further, to verify the effectiveness of these features in discriminating anomalous PD gait pattern, Support Vector Machine and Naive Bayes classifier are used as classifiers. Initial results showed that SVM with RBF kernel outshine other classifiers with 96.9% accuracy, 100% sensitivity and 95% specificity.

Vertical ground reaction force (vGRF) has been commonly used in human gait analysis making possible the study of mechanical overloads in the locomotor system. This study aimed at applying the principal component (PC) analysis and two Artificial Neural Networks (ANN), multi-layer feed forward (FF) and self organized maps (SOM), for classifying and clustering gait patterns from normal subjects (CG) and patients with lower limb fractures (FG). The vGRF from a group of 51 subjects, including 38 in CG and 13 in FG were used for PC analysis and classification. It was also tested the classification of vGRF from five subjects in a treatment group (TG) that were submitted to a physiotherapeutic treatment. Better results were obtained using four PC as inputs of the ANN, with 96% accuracy, 100% specificity and 85% sensitivity using SOM, against 92% accuracy, 100% specificity and 69% sensitivity for FF classification. After treatment, three of five subjects were classified as presenting normal vGRF.