Application of principal component analysis in vertical ground reaction force to discriminate normal and abnormal gait

An Efficient Gait Abnormality Detection Method Based on Classification

In the study of human mobility, gait analysis is a well-recognized assessment methodology. Despite its widespread use, doubts exist about its clinical utility, i.e., its potential to influence the diagnostic-therapeutic practice. Gait analysis evaluates the walking pattern (normal/abnormal) based on the gait cycle. Based on the analysis obtained, various applications can be developed in the medical, security, sports, and fitness domain to improve overall outcomes. Wearable sensors provide a convenient, efficient, and low-cost approach to gather data, while machine learning methods provide high accuracy gait feature extraction for analysis. The problem is to identify gait abnormalities and if present, subsequently identify the locations of impairments that lead to the change in gait pattern of the individual. Proper physiotherapy treatment can be provided once the location/landmark of the impairment is known correctly. In this paper, classification of multiple anatomical regions and their combination on a large scale highly imbalanced dataset is carried out. We focus on identifying 27 different locations of injury and formulate it as a multi-class classification approach. The advantage of this method is the convenience and simplicity as compared to previous methods. In our work, a benchmark is set to identify the gait disorders caused by accidental impairments at multiple anatomical regions using the GaitRec dataset. In our work, machine learning models are trained and tested on the GaitRec dataset, which provides Ground Reaction Force (GRF) data, to analyze an individual’s gait and further classify the gait abnormality (if present) at the specific lower-region portion of the body. The design and implementation of machine learning models are carried out to detect and classify the gait patterns between healthy controls and gait disorders. Finally, the efficacy of the proposed approach is showcased using various qualitative accuracy metrics. The achieved test accuracy is 96% and an F1 score of 95% is obtained in classifying various gait disorders on unseen test samples. The paper concludes by stating how machine learning models can help to detect gait abnormalities along with directions of future work.

Three decades of gait index development: A comparative review of clinical and research gait indices

BACKGROUND

A wide variety of indices have been developed to quantify gait performance markers and associate them with their respective pathologies. Indices scores have enabled better decisions regarding patient treatments and allowed for optimized monitoring of the evolution of their condition. The extensive range of human gait indices presented over the last 30 years is evaluated and summarized in this narrative literature review exploring their application in clinical and research environments.

METHODS

The analysis will explore historical and modern gait indices, focusing on the clinical efficacy with respect to their proposed pathology, age range, and associated parameter limits. Features, methods, and clinically acceptable errors are discussed while simultaneously assessing indices advantages and disadvantages. This review analyses all indices published between 1994 and February 2021 identified using the Medline, PubMed, ScienceDirect, CINAHL, EMBASE, and Google Scholar databases.

FINDINGS

A total of 30 indices were identified as noteworthy for clinical and research purposes and another 137 works were included for discussion. The indices were divided in three major groups: observational (13), instrumented (16) and hybrid (1). The instrumented indices were further sub-divided in six groups, namely kinematic- (4), spatiotemporal- (5), kinetic- (2), kinematic- and kinetic- (2), electromyographic- (1) and Inertial Measurement Unit-based indices (2).

INTERPRETATION

This work is one of the first reviews to summarize observational and instrumented gait indices, exploring their applicability in research and clinical contexts. The aim of this review is to assist members of these communities with the selection of the proper index for the group in analysis.

Principal Component Analysis of the Running Ground Reaction Forces With Different Speeds

Ground reaction force (GRF) is a key metric in biomechanical research, including parameters of loading rate (LR), first impact peak, second impact peak, and transient between first and second impact peaks in heel strike runners. The GRFs vary over time during stance. This study was aimed to investigate the variances of GRFs in rearfoot striking runners across incremental speeds. Thirty female and male runners joined the running tests on the instrumented treadmill with speeds of 2.7, 3.0, 3.3, and 3.7 m/s. The discrete parameters of vertical average loading rate in the current study are consistent with the literature findings. The principal component analysis was modeled to investigate the main variances (95%) in the GRFs over stance. The females varied in the magnitude of braking and propulsive forces (PC1, 84.93%), whereas the male runners varied in the timing of propulsion (PC1, 53.38%). The female runners dominantly varied in the transient between the first and second peaks of vertical GRF (PC1, 36.52%) and LR (PC2, 33.76%), whereas the males variated in the LR and second peak of vertical GRF (PC1, 78.69%). Knowledge reported in the current study suggested the difference of the magnitude and patterns of GRF between male and female runners across different speeds. These findings may have implications for the prevention of sex-specific running-related injuries and could be integrated with wearable signals for the in-field prediction and estimation of impact loadings and GRFs.

Estimation of the ground reaction forces from a single video camera based on the spring-like center of mass dynamics of human walking

In clinical studies, the ground reaction forces (GRFs) during walking have found being highly useful. Therefore, the force sensing shoes with small sensors and estimation methods based on kinematics from motion capture systems or inertial measurement units were proposed. Recent studies demonstrated methods of extracting GRFs from whole-body joint kinematics, which requires a significant computational load. In this study, we propose a vertical and anterior-posterior GRFs estimation method using a single camera based on the dynamic relationship between the center of mass (CoM) and the GRFs in terms of spring mechanics. The estimation method consisted of two steps: the extraction of the vertical CoM from the video clip and the conversion of the CoM information into GRFs using a walking model. From the image of the greater trochanter that is positioned near the pelvic joint, the vertical CoM was extracted. This was done after removing the artifacts by pelvic rotation and postural change of lower limbs. The parameters of a compliant bipedal walking model were tuned to best match the CoM trajectory coupled with GRFs by spring mechanics. A video camera was used to record the walking trials of five healthy young participants from the side. The walking trials was conducted at three different speeds on the instrumented treadmill; each lasted one minute long. The GRF prediction errors were approximately 9-11%, with the best matching trials found to be at a self-selected gait speed. The prediction of anterior-posterior GRF components showed a more consistent match than the vertical GRF. The results demonstrated the possibility of marker-less kinetics prediction from video images incorporating the mechanical characteristics of the CoM.

Effects of exoskeletal gait assistance on the recovery motion following tripping

Physical assistant robots improve the user’s ability to walk. However, they also potentially affect recovery motion following tripping. The assist algorithm should not interfere with the recovery motion, and should enhance the ability of the user to recover after tripping. Thus, in this study, we investigated the recovery motion affected by the assist robot after tripping. We compared the recovery motion with different reaction algorithms. Principal component analysis revealed the effects of the reaction algorithm. Correspondingly, principal components were related to the recovery motion during two steps following tripping. Specifically, the effects of the reaction algorithm were related to a principal component that represented the motion of the second step after tripping and that increased the margin of stability. Furthermore, the margin of stability became significantly large when the assistive torque was applied during the recovery motion. The result of this study suggests that the assist robot can potentially enhances the recovery motion of its user following tripping.

Human Gait Analysis Metric for Gait Retraining

The combined gait asymmetry metric (CGAM) provides a method to synthesize human gait motion. The metric is weighted to balance each parameter's effect by normalizing the data so all parameters are more equally weighted. It is designed to combine spatial, temporal, kinematic, and kinetic gait parameter asymmetries. It can also combine subsets of the different gait parameters to provide a more thorough analysis. The single number quantifying gait could assist robotic rehabilitation methods to optimize the resulting gait patterns. CGAM will help define quantitative thresholds for achievable balanced overall gait asymmetry. The study presented here compares the combined gait parameters with clinical measures such as timed up and go (TUG), six-minute walk test (6MWT), and gait velocity. The comparisons are made on gait data collected on individuals with stroke before and after twelve sessions of rehabilitation. Step length, step time, and swing time showed a strong correlation to CGAM, but the double limb support asymmetry has nearly no correlation with CGAM and ground reaction force asymmetry has a weak correlation. The CGAM scores were moderately correlated with TUG and strongly correlated to 6MWT and gait velocity.

Data reduction prior to inference: Are there consequences of comparing groups using a t-test based on principal component scores?

Researchers often use a two-step process to analyze multivariate data. First, dimensionality is reduced using a technique such as principal component analysis, followed by a group comparison using a t -test or analysis of variance. Although this practice is often discouraged, the statistical properties of this procedure are not well understood, starting with the hypothesis being tested. We suggest that this approach might be considering two distinct hypotheses, one of which is a global test of no differences in the mean vectors, and the other being a focused test of a specific linear combination where the coefficients have been estimated from the data. We study the asymptotic properties of the two-sample t -statistic for these two scenarios, assuming a nonsparse setting. We show that the size of the global test agrees with the presumed level but that the test has poor power. In contrast, the size of the focused test can be arbitrarily distorted with certain mean and covariance structures. A simple method is provided to correct the size of the focused test. Data analyses and simulations are used to illustrate the results. Recommendations on the use of this two-step method and the related use of principal components for prediction are provided.

Race Walking Ground Reaction Forces at Increasing Speeds: A Comparison with Walking and Running

Race walking has been theoretically described as a walking gait in which no flight time is allowed and high travelling speed, comparable to running (3.6–4.2 m s−1), is achieved. The aim of this study was to mechanically understand such a “hybrid gait” by analysing the ground reaction forces (GRFs) generated in a wide range of race walking speeds, while comparing them to running and walking. Fifteen athletes race-walked on an instrumented walkway (4 m) and three-dimensional GRFs were recorded at 1000 Hz. Subjects were asked to performed three self-selected speeds corresponding to a low, medium and high speed. Peak forces increased with speeds and medio-lateral and braking peaks were higher than in walking and running, whereas the vertical peaks were higher than walking but lower than running. Vertical GRF traces showed two characteristic patterns: one resembling the “M-shape” of walking and the second characterised by a first peak and a subsequent plateau. These different patterns were not related to the athletes’ performance level. The analysis of the body centre of mass trajectory, which reaches its vertical minimum at mid-stance, showed that race walking should be considered a bouncing gait regardless of the presence or absence of a flight phase.

KAVAGait: Knowledge-Assisted Visual Analytics for Clinical Gait Analysis

In 2014, more than 10 million people in the US were affected by an ambulatory disability. Thus, gait rehabilitation is a crucial part of health care systems. The quantification of human locomotion enables clinicians to describe and analyze a patient's gait performance in detail and allows them to base clinical decisions on objective data. These assessments generate a vast amount of complex data which need to be interpreted in a short time period. We conducted a design study in cooperation with gait analysis experts to develop a novel Knowledge-Assisted Visual Analytics solution for clinical Gait analysis (KAVAGait). KAVAGait allows the clinician to store and inspect complex data derived during clinical gait analysis. The system incorporates innovative and interactive visual interface concepts, which were developed based on the needs of clinicians. Additionally, an explicit knowledge store (EKS) allows externalization and storage of implicit knowledge from clinicians. It makes this information available for others, supporting the process of data inspection and clinical decision making. We validated our system by conducting expert reviews, a user study, and a case study. Results suggest that KAVAGait is able to support a clinician during clinical practice by visualizing complex gait data and providing knowledge of other clinicians.

Which osteoarthritic gait features recover following total knee replacement surgery?

Background

Gait analysis can be used to measure variations in joint function in patients with knee osteoarthritis (OA), and is useful when observing longitudinal biomechanical changes following Total Knee Replacement (TKR) surgery. The Cardiff Classifier is an objective classification tool applied previously to examine the extent of biomechanical recovery following TKR. In this study, it is further developed to reveal the salient features that contribute to recovery towards healthy function.

Methods

Gait analysis was performed on 30 patients before and after TKR surgery, and 30 healthy controls. Median TKR follow-up time was 13 months. The combined application of principal component analysis (PCA) and the Cardiff Classifier defined 18 biomechanical features that discriminated OA from healthy gait. Statistical analysis tested whether these features were affected by TKR surgery and, if so, whether they recovered to values found for the controls.

Results

The Cardiff Classifier successfully discriminated between OA and healthy gait in all 60 cases. Of the 18 discriminatory features, only six (33%) were significantly affected by surgery, including features in all three planes of the ground reaction force (p<0.001), ankle dorsiflexion moment (p<0.001), hip adduction moment (p = 0.003), and transverse hip angle (p = 0.007). All but two (89%) of these features remained significantly different to those of the control group after surgery.

Conclusions

This approach was able to discriminate gait biomechanics associated with knee OA. The ground reaction force provided the strongest discriminatory features. Despite increased gait velocity and improvements in self-reported pain and function, which would normally be clinical indicators of recovery, the majority of features were not affected by TKR surgery. This TKR cohort retained pre-operative gait patterns; reduced sagittal hip and knee moments, decreased knee flexion, increased hip flexion, and reduced hip adduction. The changes that were associated with surgery were predominantly found at the ankle and hip, rather than at the knee.

Evaluation of plantar pressures in people with hallux valgus using principal component analysis

BACKGROUND

Foot deformities are serious problems in the elderly as they increase the risk of falling.

OBJECTIVE

This study aimed to examine the relationship between foot plantar pressure during gait and hallux valgus (HV).

METHODS

Foot-pressure data were recorded using an F-scan II system. We analyzed the entire waveform of plantar pressure during gait from 37 healthy adults using principle component analysis (PCA), conducted using a 370 × 357 matrix of time-normalized plantar data of 7 areas during gait (5 gait trials × 2 (both feet) of 37 participants × 51 data points × 7 plantar areas). Two-way (plantar pressure × presence or absence of HV) analyses of variance were conducted on the principal component scores (PCSs) of principal component vectors (PCVs) 1 through 5, each of which exhibited more than 5% variance.

RESULTS

The PCA clarified that the 2nd, 3rd, and 5th PCVs (PCV 2, 3, and 5) were related to HV (p< 0.01). These PCVs exhibit a significant interaction between plantar pressure area and HV presence.

CONCLUSIONS

The larger plantar pressure of the HV group around the hallux area during walking compared with the non-HV group is a dominant difference in plantar pressure features due to HV.

Assessment of Whole Body and Local Muscle Fatigue Using Electromyography and a Perceived Exertion Scale for Squat Lifting

This research study aims at addressing the paradigm of whole body fatigue and local muscle fatigue detection for squat lifting. For this purpose, a comparison was made between perceived exertion with the heart rate and normalized mean power frequency (NMPF) of eight major muscles. The sample consisted of 25 healthy males (age: 30 ± 2.2 years). Borg’s CR-10 scale was used for perceived exertion for two segments of the body (lower and upper) and the whole body. The lower extremity of the body was observed to be dominant compared to the upper and whole body in perceived response. First mode of principal component analysis (PCA) was obtained through the covariance matrix for the eight muscles for 25 subjects for NMPF of eight muscles. The diagonal entries in the covariance matrix were observed for each muscle. The muscle with the highest absolute magnitude was observed across all the 25 subjects. The medial deltoid and the rectus femoris muscles were observed to have the highest frequency for each PCA across 25 subjects. The rectus femoris, having the highest counts in all subjects, validated that the lower extremity dominates the sense of whole body fatigue during squat lifting. The findings revealed that it is significant to take into account the relation between perceived and measured effort that can help prevent musculoskeletal disorders in repetitive occupational tasks.

Applying dimension reduction to EEG data by Principal Component Analysis reduces the quality of its subsequent Independent Component decomposition

Independent Component Analysis (ICA) has proven to be an effective data driven method for analyzing EEG data, separating signals from temporally and functionally independent brain and non-brain source processes and thereby increasing their definition. Dimension reduction by Principal Component Analysis (PCA) has often been recommended before ICA decomposition of EEG data, both to minimize the amount of required data and computation time. Here we compared ICA decompositions of fourteen 72-channel single subject EEG data sets obtained (i) after applying preliminary dimension reduction by PCA, (ii) after applying no such dimension reduction, or else (iii) applying PCA only. Reducing the data rank by PCA (even to remove only 1% of data variance) adversely affected both the numbers of dipolar independent components (ICs) and their stability under repeated decomposition. For example, decomposing a principal subspace retaining 95% of original data variance reduced the mean number of recovered 'dipolar' ICs from 30 to 10 per data set and reduced median IC stability from 90% to 76%. PCA rank reduction also decreased the numbers of near-equivalent ICs across subjects. For instance, decomposing a principal subspace retaining 95% of data variance reduced the number of subjects represented in an IC cluster accounting for frontal midline theta activity from 11 to 5. PCA rank reduction also increased uncertainty in the equivalent dipole positions and spectra of the IC brain effective sources. These results suggest that when applying ICA decomposition to EEG data, PCA rank reduction should best be avoided.

Comparing Gait with Multiple Physical Asymmetries Using Consolidated Metrics

Physical changes such as leg length discrepancy, the addition of a mass at the distal end of the leg, the use of a prosthetic, and stroke frequently result in an asymmetric gait. This paper presents a metric that can potentially serve as a benchmark to categorize and differentiate between multiple asymmetric bipedal gaits. The combined gait asymmetry metric (CGAM) is based on modified Mahalanobis distances, and it utilizes the asymmetries of gait parameters obtained from motion capture and force data recorded during human walking. The gait parameters that were used in this analysis represent spatio-temporal, kinematic, and kinetic parameters. This form of a consolidated metric will help researchers identify overall gait asymmetry by showing them if the overall gait symmetry is improving and avoid the case where one parameter's symmetry is improving while another is getting worse. The CGAM metric successfully served as a measure for overall symmetry with eleven different gait parameters and successfully showed differences among gait with multiple physical asymmetries. The results showed that mass at the distal end had a larger magnitude on overall gait asymmetry compared to leg length discrepancy. It also showed that the combined effects are varied based on the cancelation effect between gait parameters. The metric was also successful in delineating the differences of prosthetic gait and able-bodied gait at three different walking velocities.

Automatic Classification of Functional Gait Disorders

This paper proposes a comprehensive investigation of the automatic classification of functional gait disorders (GDs) based solely on ground reaction force (GRF) measurements. The aim of this study is twofold: first, to investigate the suitability of the state-of-the-art GRF parameterization techniques (representations) for the discrimination of functional GDs; and second, to provide a first performance baseline for the automated classification of functional GDs for a large-scale dataset. The utilized database comprises GRF measurements from 279 patients with GDs and data from 161 healthy controls (N). Patients were manually classified into four classes with different functional impairments associated with the "hip", "knee", "ankle", and "calcaneus". Different parameterizations are investigated: GRF parameters, global principal component analysis (PCA) based representations, and a combined representation applying PCA on GRF parameters. The discriminative power of each parameterization for different classes is investigated by linear discriminant analysis. Based on this analysis, two classification experiments are pursued: distinction between healthy and impaired gait (N versus GD) and multiclass classification between healthy gait and all four GD classes. Experiments show promising results and reveal among others that several factors, such as imbalanced class cardinalities and varying numbers of measurement sessions per patient, have a strong impact on the classification accuracy and therefore need to be taken into account. The results represent a promising first step toward the automated classification of GDs and a first performance baseline for future developments in this direction.

Tensor Decomposition of Gait Dynamics in Parkinson's Disease

OBJECTIVE

The study of gait in Parkinson's disease is important because it can provide insights into the complex neural system and physiological behaviors of the disease, of which understanding can help improve treatment and lead to effective developments of alternative neural rehabilitation programs. This paper aims to introduce an effective computational method for multichannel or multisensor data analysis of gait dynamics in Parkinson's disease.

METHOD

A model of tensor decomposition, which is a generalization of matrix-based analysis for higher dimensional analysis, is designed for differentiating multisensor time series of gait force between Parkinson's disease and healthy control cohorts.

RESULTS

Experimental results obtained from the tensor decomposition model using a PhysioNet database show several discriminating characteristics of the two cohorts, and the achievement of 100% sensitivity and 100% specificity under various cross validations.

CONCLUSION

Tensor decomposition is a useful method for the modeling and analysis of multisensor time series in patients with Parkinson's disease.

SIGNIFICANCE

Tensor-decomposition factors can be potentially used as physiological markers for Parkinson's disease, and effective features for machine learning that can provide early prediction of the disease progression.

Combined gait asymmetry metric

People with physical impairments often have asymmetric gait. To evaluate if their overall symmetry is improving during intervention, there needs to be a simple metric that can help classify gait patterns that includes multiple measures of gait asymmetry. The Combined Gait Asymmetry Metric presented here is based on the Mahalanobis distance of multiple step parameters. We tested able-bodied subjects with perturbations that involve a change in leg length, the addition of ankle weights, and a combination of both perturbations. The Mahalanobis distances are calculated from perfect symmetry to all points in the data to analyze the effects of the different perturbations. The metric demonstrates how an overall view of symmetry can give a better perspective of asymmetry than only looking at a few individual parameters. This metric is straightforward and can be extended to include large numbers of spatiotemporal, kinematic, and kinetic parameters that more completely evaluate a change in gait symmetry.

Immediate effects of EVA midsole resilience and upper shoe structure on running biomechanics: a machine learning approach

Background

Resilience of midsole material and the upper structure of the shoe are conceptual characteristics that can interfere in running biomechanics patterns. Artificial intelligence techniques can capture features from the entire waveform, adding new perspective for biomechanical analysis. This study tested the influence of shoe midsole resilience and upper structure on running kinematics and kinetics of non-professional runners by using feature selection, information gain, and artificial neural network analysis.

Methods

Twenty-seven experienced male runners (63 ± 44 km/week run) ran in four-shoe design that combined two resilience-cushioning materials (low and high) and two uppers (minimalist and structured). Kinematic data was acquired by six infrared cameras at 300 Hz, and ground reaction forces were acquired by two force plates at 1,200 Hz. We conducted a Machine Learning analysis to identify features from the complete kinematic and kinetic time series and from 42 discrete variables that had better discriminate the four shoes studied. For that analysis, we built an input data matrix of dimensions 1,080 (10 trials × 4 shoes × 27 subjects) × 1,254 (3 joints × 3 planes of movement × 101 data points + 3 vectors forces × 101 data points + 42 discrete calculated kinetic and kinematic features).

Results

The applied feature selection by information gain and artificial neural networks successfully differentiated the two resilience materials using 200(16%) biomechanical variables with an accuracy of 84.8% by detecting alterations of running biomechanics, and the two upper structures with an accuracy of 93.9%.

Discussion

The discrimination of midsole resilience resulted in lower accuracy levels than did the discrimination of the shoe uppers. In both cases, the ground reaction forces were among the 25 most relevant features. The resilience of the cushioning material caused significant effects on initial heel impact, while the effects of different uppers were distributed along the stance phase of running. Biomechanical changes due to shoe midsole resilience seemed to be subject-dependent, while those due to upper structure seemed to be subject-independent.

Variability of spatial temporal gait parameters and center of pressure displacements during gait in elderly fallers and nonfallers: A 6-month prospective study

Considering that most of the falls in elderly population arise during walking, tests derived from walking performance would be desirable for comprehensive fall risk assessment. The analysis of spatial temporal parameters and the center of pressure displacement, which represents the interaction between the human body and the ground, would be beneficial. The aim of this study was to compare spatial temporal gait parameters and their variability and the variability of the center of pressure displacement between elderly fallers and nonfallers during gait at self-selected, defined and fast speeds. A prospective study design was used. At the baseline, measurements of ground reaction force during gait at self-selected, defined and fast walking speeds by two force plates were performed. In addition, the Tinetti balance assessment tool, the Falls Efficacy Scale-International and the Activities-Specific Balance Confidence Scale were used. Mean and coefficient of variation of spatial temporal gait parameters and standard deviations of center of pressure displacement during loading response, midstance, terminal stance and preswing phases were calculated. Comparison of the fallers and nonfallers exhibited no significant difference in clinical tool, scales or spatial temporal parameters. Compared to nonfallers’ increased variability of walking speed at self-selected and defined speed, step width at fast walking speed and center of pressure displacement during preswing phase in medial-lateral directions at defined walking speed was found in fallers. However, application of the Holm-Bonferroni procedure for multiple comparisons exhibited no significant effect of group in any of the gait parameters. In general, our study did not observe an effect of group (fallers vs. nonfallers) on variability of spatial temporal parameters and center of pressure movement during gait. However, walking speed, step width as well as standard deviation of COP displacement in the medial-lateral direction during preswing exhibited a certain potential for distinguishing between elderly fallers and nonfallers.

Principal component analysis in ground reaction forces and center of pressure gait waveforms of people with transfemoral amputation

BACKGROUND

The alterations in gait pattern of people with transfemoral amputation leave them more susceptible to musculoskeletal injury. Principal component analysis is a method that reduces the amount of gait data and allows analyzing the entire waveform.

OBJECTIVES

To use the principal component analysis to compare the ground reaction force and center of pressure displacement waveforms obtained during gait between able-bodied subjects and both limbs of individuals with transfemoral amputation.

STUDY DESIGN

This is a transversal study with a convenience sample.

METHODS

We used a force plate and pressure plate to record the anterior-posterior, medial-lateral and vertical ground reaction force, and anterior-posterior and medial-lateral center of pressure positions of 12 participants with transfemoral amputation and 20 able-bodied subjects during gait. The principal component analysis was performed to compare the gait waveforms between the participants with transfemoral amputation and the able-bodied individuals.

RESULTS

The principal component analysis model explained between 74% and 93% of the data variance. In all ground reaction force and center of pressure waveforms relevant portions were identified; and always at least one principal component presented scores statistically different (p < 0.05) between the groups of participants in these relevant portions.

CONCLUSION

Principal component analysis was able to discriminate many portions of the stance phase between both lower limbs of people with transfemoral amputation compared to the able-bodied participants.

CLINICAL RELEVANCE

Principal component analysis reduced the amount of data, allowed analyzing the whole waveform, and identified specific sub-phases of gait that were different between the groups. Therefore, this approach seems to be a powerful tool to be used in gait evaluation and following the rehabilitation status of people with transfemoral amputation.

Influence of patellar position on the knee extensor mechanism in normal and crouched walking

Patella alta is common in cerebral palsy, especially in patients with crouch gait. Correction of patella alta has been advocated in the treatment of crouch, however the appropriate degree of correction and the implications for knee extensor function remain unclear. Therefore, the goal of this study was to assess the impact of patellar position on quadriceps and patellar tendon forces during normal and crouch gait. To this end, a lower extremity musculoskeletal model with a novel 12 degree of freedom knee joint was used to simulate normal gait in a healthy child, as well as mild (23 deg min knee flexion in stance), moderate (41 deg), and severe (67 deg) crouch gait in three children with cerebral palsy. The simulations revealed that quadriceps and patellar tendon forces increase dramatically with crouch, and are modulated by patellar position. For example with a normal patellar tendon position, peak patellar tendon forces were 0.7 times body weight in normal walking, but reached 2.2, 3.2 and 5.4 times body weight in mild, moderate and severe crouch. Moderate patella alta acted to reduce quadriceps and patellar tendon loads in crouch gait, due to an enhancement of the patellar tendon moment arms with alta in a flexed knee. In contrast, patella baja reduced the patellar tendon moment arm in a flexed knee and thus induced an increase in the patellar tendon loads needed to walk in crouch. Functionally, these results suggest that patella baja could also compromise knee extensor function for other flexed knee activities such as chair rise and stair climbing. The findings are important to consider when using surgical approaches for correcting patella alta in children who exhibit crouch gait patterns.

Asymmetric ground reaction forces and knee kinematics during squat after anterior cruciate ligament (ACL) reconstruction

BACKGROUND

This bilateral squat study tests whether people with anterior cruciate ligament (ACL) reconstruction have symmetric three-dimensional ground reaction forces (GRFs) and symmetric anterior-posterior (AP) translation rates of the femur with respect to the tibia when compared with healthy control subjects. We hypothesized that there would be no long-term asymmetry in knee kinematics and kinetics in ACL reconstructed subjects following surgery and rehabilitation.

METHODS

Position and GRF data were collected on eight ACL reconstructed and eight control subjects during bilateral squat. The rate of relative AP translation was determined for each subject. Principal component models were developed for each of the three GRF waveforms. Principal component scores were used to assess symmetry within the ACL reconstructed group and within the control group.

RESULTS

ACL reconstructed knees analyzed in early flexion during squat descent displayed a four-fold greater rate of change in anterior translation in the reconstructed knee relative to the contralateral side than did a similar comparison of normal knees. Differences were found between the ACL reconstructed subjects' injured and uninjured limbs for all GRFs.

CONCLUSIONS

Subjects following ACL reconstruction had asymmetric GRFs and relative rates of AP translation at an average of seven years after ACL reconstructive surgery when compared with control subjects.

CLINICAL RELEVANCE

These alterations in loading may lead to altered load distributions across the knee joint and may put some subjects at risk for future complications such as osteoarthritis.

Parkinson's disease classification using gait analysis via deterministic learning

Gait analysis plays an important role in maintaining the well-being of human mobility and health care, and is a valuable tool for obtaining quantitative information on motor deficits in Parkinson's disease (PD). In this paper, we propose a method to classify (diagnose) patients with PD and healthy control subjects using gait analysis via deterministic learning theory. The classification approach consists of two phases: a training phase and a classification phase. In the training phase, gait characteristics represented by the gait dynamics are derived from the vertical ground reaction forces under the usual and self-selected paces of the subjects. The gait dynamics underlying gait patterns of healthy controls and PD patients are locally accurately approximated by radial basis function (RBF) neural networks. The obtained knowledge of approximated gait dynamics is stored in constant RBF networks. The gait patterns of healthy controls and PD patients constitute a training set. In the classification phase, a bank of dynamical estimators is constructed for all the training gait patterns. Prior knowledge of gait dynamics represented by the constant RBF networks is embedded in the estimators. By comparing the set of estimators with a test gait pattern of a certain PD patient to be classified (diagnosed), a set of classification errors are generated. The average L₁ norms of the errors are taken as the classification measure between the dynamics of the training gait patterns and the dynamics of the test PD gait pattern according to the smallest error principle. When the gait patterns of 93 PD patients and 73 healthy controls are classified with five-fold cross-validation method, the accuracy, sensitivity and specificity of the results are 96.39%, 96.77% and 95.89%, respectively. Based on the results, it may be claimed that the features and the classifiers used in the present study could effectively separate the gait patterns between the groups of PD patients and healthy controls.

Feature Selection and Predictors of Falls with Foot Force Sensors Using KNN-Based Algorithms

The aging process may lead to the degradation of lower extremity function in the elderly population, which can restrict their daily quality of life and gradually increase the fall risk. We aimed to determine whether objective measures of physical function could predict subsequent falls. Ground reaction force (GRF) data, which was quantified by sample entropy, was collected by foot force sensors. Thirty eight subjects (23 fallers and 15 non-fallers) participated in functional movement tests, including walking and sit-to-stand (STS). A feature selection algorithm was used to select relevant features to classify the elderly into two groups: at risk and not at risk of falling down, for three KNN-based classifiers: local mean-based k-nearest neighbor (LMKNN), pseudo nearest neighbor (PNN), local mean pseudo nearest neighbor (LMPNN) classification. We compared classification performances, and achieved the best results with LMPNN, with sensitivity, specificity and accuracy all 100%. Moreover, a subset of GRFs was significantly different between the two groups via Wilcoxon rank sum test, which is compatible with the classification results. This method could potentially be used by non-experts to monitor balance and the risk of falling down in the elderly population.

Influence of wedges on lower limbs' kinematics and net joint moments during healthy elderly gait using principal component analysis

The elderly are susceptible to many disorders that alter the gait pattern and could lead to falls and reduction of mobility. One of the most applied therapeutical approaches to correct altered gait patterns is the insertion of insoles. Principal Component Analysis (PCA) is a powerful method used to reduce redundant information and it allows the comparison of the complete waveform. The purpose of this study was to verify the influence of wedges on lower limbs' net joint moment and range of motion (ROM) during the gait of healthy elderly participants using PCA. In addition, discrete values of lower limbs' peak net moment and ROM were also evaluated. 20 subjects walked with no wedges (control condition) and wearing six different wedges. The variables analyzed were the Principal Components from joint net moments and ROM in the sagittal plane in the ankle and knee and joint net moments in frontal plane in the knee. The discrete variables were peak joint net moments and ROM in sagittal plane in knee and ankle. The results showed the influence of the wedges to be clearer by analyzing through PCA methods than to use discrete parameters of gait curves, where the differences between conditions could be hidden.

OPTIMAL ESTIMATION OF TOTAL PLANTAR FORCE FOR MONITORING GAIT IN DAILY LIFE ACTIVITIES WITH LOW-PRICE INSOLE SYSTEM

This study presented optimal estimation of total plantar force with a suitable sensor layout and a reliable equation for monitoring gait in daily life activities. The plantar pressure of 10 subjects during level walking was measured by Pedar® insoles at 100 Hz for establishing models and selecting the optimal one. Four types of virtual sensors with different sizes were designed. Stepwise linear regressions were performed to reconstruct total plantar force based on each particular type of virtual sensor. 14 models were established, which met the requirements of the explained variance of the regression model and the multicollinearity of the predictors. Estimated total plantar force from each model was compared with the real data from the Pedar® insoles. According to the correlation coefficient (R) and the root mean square error divided by the peak force (RMSE/PF), the optimal model had three sensors located under the heel and metatarsal. Another four subjects were used for validating the optimal model by performing level walking, running, vertical jump-landing, stair ascending and descending. For these five common activities, the correlation was high (R > 0.970) and the error was low (RMSE/PF < 10%). Therefore, this model can accurately estimate total plantar force in daily life activities.

Effects of supramalleolar osteotomies for ankle osteoarthritis on foot kinematics and lower leg muscle activation during walking

BACKGROUND

Early stages of asymmetric ankle osteoarthritis can be treated by joint preserving supramalleolar osteotomies that surgically realign the ankle and unload degenerated cartilage. While studies have already shown pain relief and functional improvements, the effects on gait biomechanics are largely unknown. This study investigated patients' gait pattern after supramalleolar osteotomies by focusing on foot kinematics and lower leg muscle activation.

METHODS

An instrumented three-dimensional gait analysis with simultaneous electromyography of gastrocnemius medialis and lateralis, soleus, peroneus longus, and tibialis anterior muscles was performed on 12 patients with ankle osteoarthritis, seven of which were followed up 12-18months postoperatively. Additionally, seven different long-term follow-up patients (8-9years postoperatively) and 15 healthy control subjects were measured. The waveforms of the foot kinematics and muscle activation were analyzed using principal component analysis.

FINDINGS

Compared to healthy controls, principal component scores that affected the sagittal range of motion of the hindfoot and hallux were lower in all patient groups, while scores that affected the timing of the peaks in the sagittal forefoot motion were mainly altered in short-term follow-up patients. Lower principal component scores in patients with ankle osteoarthritis and short-term follow-up patients resulted in a less pronounced peak activation of gastrocnemius medialis and soleus.

INTERPRETATION

Both postoperative patient groups showed similar adaptations in their gait pattern as those observed in patients with ankle osteoarthritis. These changes are probably related to the lower ankle mobility. However, the reduced mobility seems to affect the patients' well-being less than a painful joint.

Minimum distraction gap: how much ankle joint space is enough in ankle distraction arthroplasty?

BACKGROUND

The success of ankle distraction arthroplasty relies on the separation of the tibiotalar articular surfaces.

QUESTION/PURPOSE

The purpose of this study was to find the minimum distraction gap needed to ensure that the tibiotalar joint surfaces would not contact each other with full weight-bearing while under distraction.

METHODS

Circular external fixators were mounted to nine cadaver ankle specimens. Each specimen was then placed into a custom-designed load chamber. Loads of 0, 350, and 700N were applied to the specimen. Radiographic joint space was measured and joint contact pressure was monitored under each load. The external fixator was then sequentially distracted, and the radiographic joint space was measured under the three different loads. The experiment was stopped when there was no joint contact under 700N of load. The radiographic joint space was measured and the initial (undistracted) radiographic joint space was subtracted from it yielding the distraction gap. The minimum distraction gap (mDG) that would provide total unloading was calculated.

RESULTS

The average mDG was 2.4 mm (range, 1.6 to 4.0 mm) at 700N of load, 4.4 mm (range, 3.7 to 5.8 mm) at 350N of load, and 4.9 mm (range, 3.7 to 7.0 mm) at 0N of load.

CONCLUSION

These results suggest that if the radiographic joint space of on a standing X-ray of an ankle undergoing distraction arthroplasty shows a minimum of 5.8 mm of DG, then there will be no contact between joint surfaces during full weight-bearing. Therefore, 5 mm of radiographic joint space, as recommended historically, may not be adequate to prevent contact of the articular surfaces during weight-bearing.

Principal component modeling of isokinetic moment curves for discriminating between the injured and healthy knees of unilateral ACL deficient patients

Bilateral knee strength evaluations of unilateral anterior cruciate ligament (ACL) deficient patients using isokinetic dynamometry are commonly performed in rehabilitation settings. The most frequently-used outcome measure is the peak moment value attained by the knee extensor and flexor muscle groups. However, other strength curve features may also be of clinical interest and utility. The purpose of this investigation was to identify, using Principal Component Analysis (PCA), strength curve features that explain the majority of variation between the injured and uninjured knee, and to assess the capabilities of these features to detect the presence of injury. A mixed gender cohort of 43 unilateral ACL deficient patients performed 6 continuous concentric knee extension and flexion repetitions bilaterally at 60°s(-1) and 180°s(-1) within a 90° range of motion. Moment waveforms were analyzed using PCA, and binary logistic regression was used to develop a discriminatory decision rule. For all directions and speeds, a statistically significant overall reduction in strength was noted for the involved knee in comparison to the uninvolved knee. The discriminatory decision rule yielded a specificity and sensitivity of 60.5% and 60.5%, respectively, corresponding to an accuracy of ∼62%. As such, the curve features extracted using PCA enabled only limited clinical usefulness in discerning between the ACL deficient and contra lateral, healthy knee. Improvement in discrimination capabilities may perhaps be achieved by consideration of different testing speeds and contraction modes, as well as utilization of other data analysis techniques.

Analysis of the Applicability of an Ankle-Foot Orthosis during Gait in Poststroke Patients

[Purpose] The aim of this study was to develop and assess the applicability of an experimental ankle-foot orthosis during gait in patients with hemiparesis. [Subjects and Methods] This was a noncontrolled cross-sectional study. Ten adult patients with hemiparesis but who were capable of independent gait were included in the study. Gait assessment was performed using two platforms (EMG System do Brasil), an electromyograph (EMG System do Brasil), and a video camera. The experimental orthosis consisted of a single piece that fit over the foot and 1/3 of the distal tibia and had a steel spring. [Results] There was greater activation of the rectus femoris and vastus lateralis muscles in the stance and mid-stance phases with the use of the experimental ankle-foot orthosis in comparison with the use of a polypropylene ankle-foot orthosis and no orthosis. Regarding spatial and temporal gait parameters, the individuals achieved an increase in stride length with the use of the experimental ankle-foot orthosis in comparison with the use of a polypropylene ankle-foot orthosis. [Conclusion] The results of the present study demonstrate that individuals with hemiparesis achieved an improvement in the stance and mid-stance phases of gait with the use of the experimental ankle-foot orthosis.

Classification of gait kinematics of anterior cruciate ligament reconstructed subjects using principal component analysis and regressions modelling

The aim of this study was to compare the knee kinematics of anterior cruciate ligament reconstructed (ACL-R) and healthy subjects (CG) during gait and classify the status of normality. Ten healthy and six ACL-R subjects had their gait analyzed at 60 fps. 3D knee angles were calculated and inserted into three separate matrices used to perform the principal component (PC) analysis. The scores of PCs retained in each analysis were used to calculate the standard distances (SD) of each participant in relation to the center of the CG. The PC scores of the three planes were used in a logistic regression to define normality. In the sagittal plane there was no difference between groups. In the frontal and transverse planes ACL-R subjects showed higher SD values than CG. PCs identified that ACL-R subjects showed increased adduction, internal and external rotation. All these subjects had their gait classified as abnormal by logistic regression. Therefore, in the studied ACL-R subjects the gait pattern did not return to normal levels after surgery. This may lead to degenerative injuries, as osteoarthritis, in the future.

Variations in kinematics during clinical gait analysis in stroke patients

In addition to changes in spatio-temporal and kinematic parameters, patients with stroke exhibit fear of falling as well as fatigability during gait. These changes could compromise interpretation of data from gait analysis. The aim of this study was to determine if the gait of hemiplegic patients changes significantly over successive gait trials. Forty two stroke patients and twenty healthy subjects performed 9 gait trials during a gait analysis session. The mean and variability of spatio-temporal and kinematic joint parameters were analyzed during 3 groups of consecutive gait trials (1-3, 4-6 and 7-9). Principal component analysis was used to reduce the number of variables from the joint kinematic waveforms and to identify the parts of the gait cycle which changed during the gait analysis session. The results showed that i) spontaneous gait velocity and the other spatio-temporal parameters significantly increased, and ii) gait variability decreased, over the last 6 gait trials compared to the first 3, for hemiplegic patients but not healthy subjects. Principal component analysis revealed changes in the sagittal waveforms of the hip, knee and ankle for hemiplegic patients after the first 3 gait trials. These results suggest that at the beginning of the gait analysis session, stroke patients exhibited phase of adaptation,characterized by a "cautious gait" but no fatigue was observed.

IDENTIFICATION OF KNEE FRONTAL PLANE KINEMATIC PATTERNS IN NORMAL GAIT BY PRINCIPAL COMPONENT ANALYSIS

The purpose of this study was to identify meaningful gait patterns in knee frontal plane kinematics from a large population of asymptomatic individuals. The proposed method used principal component analysis (PCA). It first reduced the data dimensionality, without loss of relevant information, by projecting the original kinematic data onto a subspace of significant principal components (PCs). This was followed by a discriminant model to separate the individuals' gait into homogeneous groups. Four descriptive gait patterns were identified and validated by clustering silhouette width and statistical hypothesis testing. The first pattern was close to neutral during the stance phase and in adduction during the swing phase (Cluster 1). The second pattern was in abduction during the stance phase and tends into adduction during the swing phase (Cluster 2). The third pattern was close to neutral during the stance phase and in abduction during the swing phase (Cluster 3) and the fourth was in abduction during both the stance and the swing phase (Cluster 4).

Analyzing individual and group differences in multijoint multiwaveform gait data using the Parafac2 model

Locomotion research often involves analyzing multiwaveform data (e.g., velocities, accelerations, etc.) from various body locations (e.g., knees, ankles, etc.) of several subjects. Therefore, some multivariate technique such as principal component analysis is often used to examine interrelationships between the many correlated waveforms. Despite its extensive use in locomotion research, principal component analysis is for two-mode data, whereas locomotion data are typically collected in higher mode form. In this paper, we present the benefits of analyzing four-mode locomotion data (subjects × time × joints × waveforms) using the Parafac2 model, which is a component model designed for analyzing variation in multimode data. Using bilateral hip, knee, and ankle angular displacement, velocity, and acceleration waveforms, we demonstrate Parafac2's ability to produce interpretable components describing (i) the fundamental patterns of variation in lower limb angular kinematics during healthy walking and (ii) the fundamental differences between normal and atypical subjects' multijoint multiwaveform locomotive patterns. Also, we illustrate how Parafac2 makes it possible to determine which waveforms best characterize the individual and/or group differences captured by each component. Our results indicate that different waveforms should be used for different purposes, confirming the need for the holistic analysis of multijoint multiwaveform locomotion data, particularly when investigating atypical motion patterns.

Gait patterns of asymmetric ankle osteoarthritis patients

BACKGROUND

In early stages, ankle osteoarthritis is often asymmetric with only partially degenerated joint surfaces. There is only limited knowledge on the effect of asymmetric ankle osteoarthritis on the patients' gait patterns. Therefore, the aim of this study was to characterize kinematic and kinetic changes compared to healthy adults.

METHODS

Instrumented gait analysis was performed in eight asymmetric ankle osteoarthritis patients and 15 healthy controls. Beside conventional gait analysis methods, principal component analysis was used to analyze temporal progression of the most important variables: hindfoot dorsiflexion angle and vertical ground reaction force.

FINDINGS

Asymmetric ankle osteoarthritis patients had a lower hindfoot dorsiflexion and rotation range of motion as well as reduced peak ground reaction forces and peak kinetic values. Principal component analysis revealed that for both the hindfoot dorsiflexion angle and the vertical ground reaction force those principal component vectors affecting the amplitudes had significantly lower principal component scores in patients than in controls. The use of the principal component scores for classification with a linear support vector machine resulted in a high recognition rate of 97.8% for the discrimination between the affected leg and the healthy controls.

INTERPRETATION

Patients with asymmetric ankle osteoarthritis suffer from substantial pathological kinematic and kinetic gait changes. Principal component analysis combined with a linear support vector machine could successfully be used to temporally quantify and classify asymmetric ankle osteoarthritis gait patterns. This study therefore helps to understand the pathomechanism of early stage ankle osteoarthritis from a biomechanical view.

Long-term evaluation of gait initiation in six Parkinson's disease patients with bilateral subthalamic stimulation

Defined as the transient state between standing and walking, gait initiation is negatively affected in Parkinson's disease (PD), which often results in significant disability. Although deep brain stimulation (DBS) is the most common surgical procedure for PD, the long-term effects of DBS on gait initiation are not well studied. The present study evaluated the long-term effects of subthalamic nucleus (STN) DBS on the preparation phase of gait initiation using principal component (PC) analysis. Six patients with PD who had undergone STN DBS and 24 healthy control subjects were evaluated. PD subjects were assessed 11.3±10.3 (P1) and 78.9±10.6 (P2) months after surgery. PD subjects were tested with STN DBS in two conditions: without medication and with medication. PC analysis was applied separately for the vertical, anterior-posterior and medial-lateral components of ground reaction force (GRF) recorded during gait initiation. Three PC scores were chosen by the scree test for each GRF component and all these PC scores were used for calculating a standard distance between healthy controls and PD subjects. The Friedman test showed a significant difference in standard distance among conditions (P=0.004), with the post-hoc test recognizing differences among P1 conditions and P2 medication-on condition. The eigenvector loading factors pointed to major differences between PD conditions surrounding the maximum amplitude of vertical and anterior-posterior GRF. For the studied sample, all distances increased in the follow-up evaluation (P2) with and without medications, indicating a worsening in gait initiation after seven years.

Use of gait summary measures with lower limb amputees

Gait summary measures have been developed as a convenient method to communicate overall gait pathology. These measures are primarily used in the context of paediatric cerebral palsy and their use remains largely untested in other disability groups. This study assessed the suitability of gait summary measures for use with lower limb amputees. Modified (m) versions of three published gait summary measures were investigated - the Gillette Gait Index (mGGI), the Gait Deviation Index (mGDI) and the Gait Profile Score (mGPS) in conjunction with the Movement Analysis Profile (MAP). Twenty unilateral lower limb amputees underwent three-dimensional gait analysis. All measures reported significant differences between levels of amputation on the prosthetic limb. The mGGI and mGPS detected significant differences between the levels of amputation on the intact side, but the mGDI did not. All gait summary measures were moderately to strongly correlated with leg-length normalised self-selected walking speed and strong correlations were reported between all measures. The MAP exposed common strategies in amputee gait and showed that sagittal hip and knee kinematics contributed predominantly to overall gait deviation in this population group. The mGGI, mGDI and mGPS identified, quantified and stratified gait pathology, indicating that any of the gait measures investigated in this study can be applied as outcome measures in research and case management in lower limb amputees.

Human gait classification after lower limb fracture using Artificial Neural Networks and principal component analysis

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.