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

Transfemoral prosthetic simulators versus amputees: ground reaction forces and spatio-temporal parameters in gait

This study aimed to compare the ground reaction forces (GRFs) and spatio-temporal parameters as well as their asymmetry ratios in gait between individuals wearing a transfemoral prosthetic simulator (TFSim) and individuals with unilateral transfemoral amputation (TFAmp) across a range of walking speeds (2.0-5.5 km h-1). The study recruited 10 non-disabled individuals using TFSim and 10 individuals with unilateral TFAmp using a transfemoral prosthesis. Data were collected using an instrumented treadmill with built-in force plates, and subsequently, the GRFs and spatio-temporal parameters, as well as their asymmetry ratios, were analysed. When comparing the TFSim and TFAmp groups, no significant differences were found among the gait parameters and asymmetry ratios of all tested metrics except the vertical GRFs. The TFSim may not realistically reproduce the vertical GRFs during the weight acceptance and push-off phases. The structural and functional variations in prosthetic limbs and components between the TFSim and TFAmp groups may be primary contributors to the difference in the vertical GRFs. These results suggest that TFSim might be able to emulate the gait of individuals with TFAmp regarding the majority of spatio-temporal and GRF parameters. However, the vertical GRFs of TFSim should be interpreted with caution.

Ground reaction forces during double limb stances while walking in individuals with unilateral transfemoral amputation

The asymmetrical gait of individuals with unilateral transfemoral amputation has been well documented. However, there is not a wealth of investigation into asymmetries during the double limb stance depending on whether the intact or prosthetic limb is leading. The first aim of this study was to compare ground reaction forces during the double limb stance of individuals with unilateral transfemoral amputation depending on whether their intact (initial double limb stance) or prosthetic (terminal double limb stance) limb was leading. The second aim of this study was to compare the asymmetry ratio of ground reaction forces during the double limb stance between individuals with and without unilateral transfemoral amputation. Thirty individuals, fifteen with unilateral transfemoral amputation and fifteen who were able-bodied, were recruited for this study. Each individual walked on an instrumented treadmill for 30 s at eight different speeds, ranging from 2.0 km/h to 5.5 km/h with .5 km/h increments. Ground reaction force parameters, temporal parameters, and asymmetry ratios of all parameters were computed from the data collected. The appropriate statistical analyses of all data based on normality were conducted to investigate the aims of this study. Significant main effects of speed, double limb stance, and their interactions were found for most parameters (p < .01 or p < .05). Individuals with unilateral transfemoral amputation spent a longer duration in terminal double limb stance than initial double limb stance at all tested speeds. They also experienced significantly higher peak vertical ground reaction force during initial double limb stance compared to terminal double limb stance with increasing walking speed. However, during terminal double limb stance, higher anteroposterior ground reaction force at initial contact was found when compared to initial double limb stance. Significant differences between individuals with unilateral transfemoral amputation and able-bodied individuals were found in asymmetry ratios for peak vertical ground reaction force, anteroposterior ground reaction force, anteroposterior shear, and mediolateral shear at all tested speeds. Asymmetrical loading persists in individuals with unilateral transfemoral amputation during double limb stance. Increasing walking speed increased ground reaction force loading asymmetries, which may make individuals with unilateral transfemoral amputation more susceptible to knee osteoarthritis or other musculoskeletal disorders. Further study is necessary to develop ideal gait strategies for the minimization of gait asymmetry in individuals with unilateral transfemoral amputation.

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.

Effects of step frequency during running on the magnitude and symmetry of ground reaction forces in individuals with a transfemoral amputation

BACKGROUND

Individuals with unilateral transfemoral amputation are prone to developing health conditions such as knee osteoarthritis, caused by additional loading on the intact limb. Such individuals who can run again may be at higher risk due to higher ground reaction forces (GRFs) as well as asymmetric gait patterns. The two aims of this study were to investigate manipulating step frequency as a method to reduce GRFs and its effect on asymmetric gait patterns in individuals with unilateral transfemoral amputation while running.

METHODS

This is a cross-sectional study. Nine experienced track and field athletes with unilateral transfemoral amputation were recruited for this study. After calculation of each participant's preferred step frequency, each individual ran on an instrumented treadmill for 20 s at nine different metronome frequencies ranging from - 20% to + 20% of the preferred frequency in increments of 5% with the help of a metronome. From the data collected, spatiotemporal parameters, three components of peak GRFs, and the components of GRF impulses were computed. The asymmetry ratio of all parameters was also calculated. Statistical analyses of all data were conducted with appropriate tools based on normality analysis to investigate the main effects of step frequency. For parameters with significant main effects, linear regression analyses were further conducted for each limb.

RESULTS

Significant main effects of step frequency were found in multiple parameters (P < 0.01). Both peak GRF and GRF impulse parameters that demonstrated significant main effects tended towards decreasing magnitude with increasing step frequency. Peak vertical GRF in particular demonstrated the most symmetric values between the limbs from - 5% to 0% metronome frequency. All parameters that demonstrated significant effects in asymmetry ratio became more asymmetric with increasing step frequency.

CONCLUSIONS

For runners with a unilateral transfemoral amputation, increasing step frequency is a viable method to decrease the magnitude of GRFs. However, with the increase of step frequency, further asymmetry in gait is observed. The relationships between step frequency, GRFs, and the asymmetry ratio in gait may provide insight into the training of runners with unilateral transfemoral amputation for the prevention of injury.

Dimensionality Reduction of Human Gait for Prosthetic Control

We seek to use dimensionality reduction to simplify the difficult task of controlling a lower limb prosthesis. Though many techniques for dimensionality reduction have been described, it is not clear which is the most appropriate for human gait data. In this study, we first compare how Principal Component Analysis (PCA) and an autoencoder on poses (Pose-AE) transform human kinematics data during flat ground and stair walking. Second, we compare the performance of PCA, Pose-AE and a new autoencoder trained on full human movement trajectories (Move-AE) in order to capture the time varying properties of gait. We compare these methods for both movement classification and identifying the individual. These are key capabilities for identifying useful data representations for prosthetic control. We first find that Pose-AE outperforms PCA on dimensionality reduction by achieving a higher Variance Accounted For (VAF) across flat ground walking data, stairs data, and undirected natural movements. We then find in our second task that Move-AE significantly outperforms both PCA and Pose-AE on movement classification and individual identification tasks. This suggests the autoencoder is more suitable than PCA for dimensionality reduction of human gait, and can be used to encode useful representations of entire movements to facilitate prosthetic control tasks.

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.

Unilateral above-knee amputees achieve symmetric mediolateral ground reaction impulse in walking using an asymmetric gait strategy

The ability to sustain steady straight-ahead walking is one goal of gait rehabilitation for individuals with unilateral above-knee (UAK) amputation. Despite the morphological and musculoskeletal asymmetry resulting from unilateral limb loss, the mediolateral ground-reaction-impulse (GRI) should be counterbalanced between the affected and unaffected limbs during straight-ahead walking. Therefore, we investigated the strategies of mediolateral ground-reaction-force (GRF) generation adopted by UAK prosthesis users walking along a straight path. GRFs of 15 participants with UAK amputation were measured during straight-ahead walking. Then, the mediolateral GRI, stance time, and mean mediolateral GRF during the stance phase of the affected and unaffected limbs were compared. To better understand the GRF generation strategy, statistical-parametric-mapping (SPM) was applied to assess the phase-dependent difference of the mediolateral GRFs between two limbs. The results showed that UAK prosthesis users can achieve symmetric mediolateral GRI during straight-ahead walking by adopting an asymmetric gait strategy: shorter stance time and higher mean mediolateral GRF over the stance phase for the affected than for the unaffected limb. In addition, the analysis using SPM revealed that the affected limb generates a higher mean medial GRF component than the unaffected limb, especially during the single-support phase. Thus, a higher medial GRF during the single-support phase of the affected limb may allow UAK prosthesis users to achieve mediolateral GRI that are similar to those of the unaffected limb. Further insights on these mechanics may serve as guidelines on the improved design of prosthetic devices and the rehabilitation needs of UAK prosthesis users.

Interpretability of Input Representations for Gait Classification in Patients after Total Hip Arthroplasty

Many machine learning models show black box characteristics and, therefore, a lack of transparency, interpretability, and trustworthiness. This strongly limits their practical application in clinical contexts. For overcoming these limitations, Explainable Artificial Intelligence (XAI) has shown promising results. The current study examined the influence of different input representations on a trained model's accuracy, interpretability, as well as clinical relevancy using XAI methods. The gait of 27 healthy subjects and 20 subjects after total hip arthroplasty (THA) was recorded with an inertial measurement unit (IMU)-based system. Three different input representations were used for classification. Local Interpretable Model-Agnostic Explanations (LIME) was used for model interpretation. The best accuracy was achieved with automatically extracted features (mean accuracy Macc = 100%), followed by features based on simple descriptive statistics (Macc = 97.38%) and waveform data (Macc = 95.88%). Globally seen, sagittal movement of the hip, knee, and pelvis as well as transversal movement of the ankle were especially important for this specific classification task. The current work shows that the type of input representation crucially determines interpretability as well as clinical relevance. A combined approach using different forms of representations seems advantageous. The results might assist physicians and therapists finding and addressing individual pathologic gait patterns.

Input representations and classification strategies for automated human gait analysis

BACKGROUND

Quantitative gait analysis produces a vast amount of data, which can be difficult to analyze. Automated gait classification based on machine learning techniques bear the potential to support clinicians in comprehending these complex data. Even though these techniques are already frequently used in the scientific community, there is no clear consensus on how the data need to be preprocessed and arranged to assure optimal classification accuracy outcomes.

RESEARCH QUESTION

Is there an optimal data aggregation and preprocessing workflow to optimize classification accuracy outcomes?

METHODS

Based on our previous work on automated classification of ground reaction force (GRF) data, a sequential setup was followed: firstly, several aggregation methods - early fusion and late fusion - were compared, and secondly, based on the best aggregation method identified, the expressiveness of different combinations of signal representations was investigated. The employed dataset included data from 910 subjects, with four gait disorder classes and one healthy control group. The machine learning pipeline comprised principle component analysis (PCA), z-standardization and a support vector machine (SVM).

RESULTS

The late fusion aggregation, i.e., utilizing majority voting on the classifier's predictions, performed best. In addition, the use of derived signal representations (relative changes and signal differences) seems to be advantageous as well.

SIGNIFICANCE

Our results indicate that great caution is needed when data preprocessing and aggregation methods are selected, as these can have an impact on classification accuracies. These results shall serve future studies as a guideline for the choice of data aggregation and preprocessing techniques to be employed.

Analysis of Internal Knee Forces Allows for the Prediction of Rupture Events in a Clinically Relevant Model of Anterior Cruciate Ligament Injuries

Background:

A recently developed mechanical impact simulator induced an anterior cruciate ligament (ACL) rupture via the application of a combination of inverse dynamics–based knee abduction moment (KAM), anterior tibial shear force (ATS), and internal tibial rotation moment with impulsive compression in a cohort of cadaveric limbs. However, there remains an opportunity to further define the interaction of internal forces and moments at the knee and their respective influence on injury events.

Purpose:

To identify the influence of internal knee loads on an ACL injury event using a cadaveric impact simulator.

Study Design:

Controlled laboratory study.

Methods:

Drop-landing simulations were performed and analyzed on 30 fresh-frozen cadaveric knees with a validated mechanical impact simulator. Internal forces and moments at the knee joint center were calculated using data from a 6-axis load cell recorded on the femur during testing. Kinetic data from a total of 1083 trials that included 30 ACL injury trials were used as inputs for principal component (PC) analysis to identify the most critical features of loading waveforms. Logistic regression analysis with a stepwise selection was used to select the PCs that predicted an ACL injury. Injurious waveforms were reconstructed with selected PCs in logistic regression analysis.

Results:

A total of 3 PCs were selected in logistic regression analysis that developed a significant model (P < .001). The external loading of KAM was highly correlated with PC1 (ρ < –0.8; P < .001), which explained the majority (>69%) of the injurious waveforms reconstructed with the 3 selected PCs. The injurious waveforms demonstrated a larger internal knee adduction moment and lateral tibial force. After the ACL was ruptured, decreased posterior tibial force was observed in injury trials.

Conclusion:

These findings give us a better understanding of ACL injury mechanisms using 6-axis kinetics from an in vitro simulator. An ACL rupture was correlated with an internal knee adduction moment (external KAM) and was augmented by ATS and lateral tibial force induced by an impact, which distorted the ACL insertion orientation.

Clinical Relevance:

The ACL injury mechanism explained in this study may help target injury prevention programs to decrease injurious knee loading (KAM, ATS, and lateral tibial force) during landing tasks.

Providing low-dimensional feedback of a high-dimensional movement allows for improved performance of a skilled walking task

Learning a skilled movement often requires changing multiple dimensions of movement in a coordinated manner. Serial training is one common approach to learning a new movement pattern, where each feature is learned in isolation from the others. Once one feature is learned, we move on to the next. However, when learning a complex movement pattern, serial training is not only laborious but can also be ineffective. Often, movement features are linked such that they cannot simply be added together as we progress through training. Thus, the ability to learn multiple features in parallel could make training faster and more effective. When using visual feedback as the tool for changing movement, however, such parallel training may increase the attentional load of training and impair performance. Here, we developed a novel visual feedback system that uses principal component analysis to weight four features of movement to create a simple one-dimensional 'summary' of performance. We used this feedback to teach healthy, young participants a modified walking pattern and compared their performance to those who received four concurrent streams of visual information to learn the same goal walking pattern. We demonstrated that those who used the principal component-based visual feedback improved their performance faster and to a greater extent compared to those who received concurrent feedback of all features. These results suggest that our novel principal component-based visual feedback provides a method for altering multiple features of movement toward a prescribed goal in an intuitive, low-dimensional manner.

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.

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.

Optimisation of a machine learning algorithm in human locomotion using principal component and discriminant function analyses

Assessment methods in human locomotion often involve the description of normalised graphical profiles and/or the extraction of discrete variables. Whilst useful, these approaches may not represent the full complexity of gait data. Multivariate statistical methods, such as Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA), have been adopted since they have the potential to overcome these data handling issues. The aim of the current study was to develop and optimise a specific machine learning algorithm for processing human locomotion data. Twenty participants ran at a self-selected speed across a 15m runway in barefoot and shod conditions. Ground reaction forces (BW) and kinematics were measured at 1000 Hz and 100 Hz, respectively from which joint angles (°), joint moments (N.m.kg-1) and joint powers (W.kg-1) for the hip, knee and ankle joints were calculated in all three anatomical planes. Using PCA and DFA, power spectra of the kinematic and kinetic variables were used as a training database for the development of a machine learning algorithm. All possible combinations of 10 out of 20 participants were explored to find the iteration of individuals that would optimise the machine learning algorithm. The results showed that the algorithm was able to successfully predict whether a participant ran shod or barefoot in 93.5% of cases. To the authors' knowledge, this is the first study to optimise the development of a machine learning algorithm.