Automatic Classification of Functional Gait Disorders

A stacked autoencoder-based aid system for severity degree classification of knee ligament rupture

BACKGROUND

Knee ligament rupture is one of the most common injuries, but the diagnosis of its severity tends to require the use of complex methods and analyses that are not always available to patients.

AIM

The objective of this research is the investigation and development of a diagnostic aid system to analyze and determine patterns that characterize the presence of the injury and its degree of severity.

METHODS

Implement a novel proposal of a framework based on stacked auto-encoder (SAE) for ground reaction force (GRF) signals analysis, coming from the GaitRec database. Analysis of the raw data is used to determine the main features that allow us to diagnose the presence of a knee ligament rupture and classify its severity as high, mid or mild.

RESULTS

The process is divided into two stages to determine the presence of the lesion and, if necessary, evaluate variations in features to classify the degree of severity as high, mid, and mild. The framework presents an accuracy of 87 % and a F1-Score of 90 % for detecting ligament rupture and an accuracy of 86.5 % and a F1-Score of 87 % for classifying severity.

CONCLUSION

This new methodology aims to demonstrate the potential of SAE in physiotherapy applications as an evaluation and diagnostic tool, identifying irregularities associated with ligament rupture and its degree of severity, thus providing updated information to the specialist during the rehabilitation process.

Human Pose Estimation for Clinical Analysis of Gait Pathologies

Gait analysis serves as a critical diagnostic tool for identifying neurologic and musculoskeletal damage. Traditional manual analysis of motion data, however, is labor-intensive and heavily reliant on the expertise and judgment of the therapist. This study introduces a binary classification method for the quantitative assessment of gait impairments, specifically focusing on Duchenne muscular dystrophy (DMD), a prevalent and fatal neuromuscular genetic disorder. The research compares spatiotemporal and sagittal kinematic gait features derived from 2D and 3D human pose estimation trajectories against concurrently recorded 3D motion capture (MoCap) data from healthy children. The proposed model leverages a novel benchmark dataset, collected from YouTube and publicly available datasets of their typically developed peers, to extract time-distance variables (e.g. speed, step length, stride time, and cadence) and sagittal joint angles of the lower extremity (e.g. hip, knee, and knee flexion angles). Machine learning and deep learning techniques are employed to discern patterns that can identify children exhibiting DMD gait disturbances. While the current model is capable of distinguishing between healthy subjects and those with DMD, it does not specifically differentiate between DMD patients and patients with other gait impairments. Experimental results validate the efficacy of our cost-effective method, which relies on recorded RGB video, in detecting gait abnormalities, achieving a prediction accuracy of 96.2% for Support Vector Machine (SVM) and 97% for the deep network.

A scoping review of applications of artificial intelligence in kinematics and kinetics of ankle sprains - current state-of-the-art and future prospects

BACKGROUND

Despite the existence of evidence-based rehabilitation strategies that address biomechanical deficits, the persistence of recurrent ankle problems in 70% of patients with acute ankle sprains highlights the unresolved nature of this issue. Artificial intelligence (AI) emerges as a promising tool to identify definitive predictors for ankle sprains. This paper aims to summarize the use of AI in investigating the ankle biomechanics of healthy and subjects with ankle sprains.

METHODS

Articles published between 2010 and 2023 were searched from five electronic databases. 59 papers were included for analysis with regards to: i). types of motion tested (functional vs. purposeful ankle movement); ii) types of biomechanical parameters measured (kinetic vs kinematic); iii) types of sensor systems used (lab-based vs field-based); and, iv) AI techniques used.

FINDINGS

Most studies (83.1%) examined biomechanics during functional motion. Single kinematic parameter, specifically ankle range of motion, could obtain accuracy up to 100% in identifying injury status. Wearable sensor exhibited high reliability for use in both laboratory and on-field/clinical settings. AI algorithms primarily utilized electromyography and joint angle information as input data. Support vector machine was the most used supervised learning algorithm (18.64%), while artificial neural network demonstrated the highest accuracy in eight studies.

INTERPRETATIONS

The potential for remote patient monitoring is evident with the adoption of field-based devices. Nevertheless, AI-based sensors are underutilized in detecting ankle motions at risk of sprain. We identify three key challenges: sensor designs, the controllability of AI models, and the integration of AI-sensor models, providing valuable insights for future research.

A new method applied for explaining the landing patterns: Interpretability analysis of machine learning

As one of many fundamental sports techniques, the landing maneuver is also frequently used in clinical injury screening and diagnosis. However, the landing patterns are different under different constraints, which will cause great difficulties for clinical experts in clinical diagnosis. Machine learning (ML) have been very successful in solving a variety of clinical diagnosis tasks, but they all have the disadvantage of being black boxes and rarely provide and explain useful information about the reasons for making a particular decision. The current work validates the feasibility of applying an explainable ML (XML) model constructed by Layer-wise Relevance Propagation (LRP) for landing pattern recognition in clinical biomechanics. This study collected 560 groups landing data. By incorporating these landing data into the XML model as input signals, the prediction results were interpreted based on the relevance score (RS) derived from LRP. The interpretation obtained from XML was evaluated comprehensively from the statistical perspective based on Statistical Parametric Mapping (SPM) and Effect Size. The RS has excellent statistical characteristics in the interpretation of landing patterns between classes, and also conforms to the clinical characteristics of landing pattern recognition. The current work highlights the applicability of XML methods that can not only satisfy the traditional decision problem between classes, but also largely solve the lack of transparency in landing pattern recognition. We provide a feasible framework for realizing interpretability of ML decision results in landing analysis, providing a methodological reference and solid foundation for future clinical diagnosis and biomechanical analysis.

Hybrid Deep Neural Network Framework Combining Skeleton and Gait Features for Pathological Gait Recognition

Human skeleton data obtained using a depth camera have been used for pathological gait recognition to support doctor or physician diagnosis decisions. Most studies for skeleton-based pathological gait recognition have used either raw skeleton sequences directly or gait features, such as gait parameters and joint angles, extracted from raw skeleton sequences. We hypothesize that using skeleton, joint angles, and gait parameters together can improve recognition performance. This study aims to develop a deep neural network model that effectively combines different types of input data. We propose a hybrid deep neural network framework composed of a graph convolutional network, recurrent neural network, and artificial neural network to effectively encode skeleton sequences, joint angle sequences, and gait parameters, respectively. The features extracted from three different input data types are fused and fed into the final classification layer. We evaluate the proposed model on two different skeleton datasets (a simulated pathological gait dataset and a vestibular disorder gait dataset) that were collected using an Azure Kinect. The proposed model, with multiple types of input, improved the pathological gait recognition performance compared to single input models on both datasets. Furthermore, it achieved the best performance among the state-of-the-art models for skeleton-based action recognition.

Empowering Lower Limb Disorder Identification through PoseNet and Artificial Intelligence

A novel approach is presented in this study for the classification of lower limb disorders, with a specific emphasis on the knee, hip, and ankle. The research employs gait analysis and the extraction of PoseNet features from video data in order to effectively identify and categorize these disorders. The PoseNet algorithm facilitates the extraction of key body joint movements and positions from videos in a non-invasive and user-friendly manner, thereby offering a comprehensive representation of lower limb movements. The features that are extracted are subsequently standardized and employed as inputs for a range of machine learning algorithms, such as Random Forest, Extra Tree Classifier, Multilayer Perceptron, Artificial Neural Networks, and Convolutional Neural Networks. The models undergo training and testing processes using a dataset consisting of 174 real patients and normal individuals collected at the Tehsil Headquarter Hospital Sadiq Abad. The evaluation of their performance is conducted through the utilization of K-fold cross-validation. The findings exhibit a notable level of accuracy and precision in the classification of various lower limb disorders. Notably, the Artificial Neural Networks model achieves the highest accuracy rate of 98.84%. The proposed methodology exhibits potential in enhancing the diagnosis and treatment planning of lower limb disorders. It presents a non-invasive and efficient method of analyzing gait patterns and identifying particular conditions.

GaitRec-Net: A Deep Neural Network for Gait Disorder Detection Using Ground Reaction Force

Walking (gait) irregularities and abnormalities are predictors and symptoms of disorder and disability. In the past, elaborate video (camera-based) systems, pressure mats, or a mix of the two has been used in clinical settings to monitor and evaluate gait. This article presents an artificial intelligence-based comprehensive investigation of ground reaction force (GRF) pattern to classify the healthy control and gait disorders using the large-scale ground reaction force. The used dataset comprised GRF measurements from different patients. The article includes machine learning- and deep learning-based models to classify healthy and gait disorder patients using ground reaction force. A deep learning-based architecture GaitRec-Net is proposed for this classification. The classification results were evaluated using various metrics, and each experiment was analysed using a fivefold cross-validation approach. Compared to machine learning classifiers, the proposed deep learning model is found better for feature extraction resulting in high accuracy of classification. As a result, the proposed framework presents a promising step in the direction of automatic categorization of abnormal gait pattern.

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.

Behavioral Ethics Ecologies of Human-Artificial Intelligence Systems

Historically, evolution of behaviors often took place in environments that changed little over millennia. By contrast, today, rapid changes to behaviors and environments come from the introduction of artificial intelligence (AI) and the infrastructures that facilitate its application. Behavioral ethics is concerned with how interactions between individuals and their environments can lead people to questionable decisions and dubious actions. For example, interactions between an individual’s self-regulatory resource depletion and organizational pressure to take non-ethical actions. In this paper, four fundamental questions of behavioral ecology are applied to analyze human behavioral ethics in human–AI systems. These four questions are concerned with assessing the function of behavioral traits, how behavioral traits evolve in populations, what are the mechanisms of behavioral traits, and how they can differ among different individuals. These four fundamental behavioral ecology questions are applied in analysis of human behavioral ethics in human–AI systems. This is achieved through reference to vehicle navigation systems and healthcare diagnostic systems, which are enabled by AI. Overall, the paper provides two main contributions. First, behavioral ecology analysis of behavioral ethics. Second, application of behavioral ecology questions to identify opportunities and challenges for ethical human–AI systems.

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.

Analysis of the stance phase of the gait cycle in Parkinson's disease and its potency for Parkinson's disease discrimination

In this study, using vertical ground reaction force (VGRF) data and focusing on the stance phase of the gait cycle, the effect of Parkinson's disease (PD) on gait was investigated. The used dataset consisted of 93 PD and 72 healthy individuals. Multiple comparisons correction ANOVA test and student t-test were used for statistical analyses. Results showed that a longer stance duration with a larger VGRF peak value (p < 0.05) was observed for PD patients during the stance phase. In addition, the VGRF peak value was delayed and blunted in PD cases compared with healthy individuals. These results indicated more time and effort for PD patients for posture stabilization during the stance phase. The time delay for different locations of the foot sole to contact the ground during the stance phase indicated that PD patients might use a different strategy for maintaining their body stability compared with healthy individuals. Although the VGRF time-domain pattern during the stance phase in PD was similar to healthy conditions, its local characteristics like duration and peak value differed significantly. The classification analysis based on the VGRF time-domain extracted features during the stance phase obtained PD recognition with accuracy, sensitivity and specificity of 90.82%, 88.63% and 82.56%, respectively.

Remote Gait Type Classification System Using Markerless 2D Video

Several pathologies can alter the way people walk, i.e., their gait. Gait analysis can be used to detect such alterations and, therefore, help diagnose certain pathologies or assess people's health and recovery. Simple vision-based systems have a considerable potential in this area, as they allow the capture of gait in unconstrained environments, such as at home or in a clinic, while the required computations can be done remotely. State-of-the-art vision-based systems for gait analysis use deep learning strategies, thus requiring a large amount of data for training. However, to the best of our knowledge, the largest publicly available pathological gait dataset contains only 10 subjects, simulating five types of gait. This paper presents a new dataset, GAIT-IT, captured from 21 subjects simulating five types of gait, at two severity levels. The dataset is recorded in a professional studio, making the sequences free of background camouflage, variations in illumination and other visual artifacts. The dataset is used to train a novel automatic gait analysis system. Compared to the state-of-the-art, the proposed system achieves a drastic reduction in the number of trainable parameters, memory requirements and execution times, while the classification accuracy is on par with the state-of-the-art. Recognizing the importance of remote healthcare, the proposed automatic gait analysis system is integrated with a prototype web application. This prototype is presently hosted in a private network, and after further tests and development it will allow people to upload a video of them walking and execute a web service that classifies their gait. The web application has a user-friendly interface usable by healthcare professionals or by laypersons. The application also makes an association between the identified type of gait and potential gait pathologies that exhibit the identified characteristics.

Gutenberg Gait Database, a ground reaction force database of level overground walking in healthy individuals

The Gutenberg Gait Database comprises data of 350 healthy individuals recorded in our laboratory over the past seven years. The database contains ground reaction force (GRF) and center of pressure (COP) data of two consecutive steps measured - by two force plates embedded in the ground - during level overground walking at self-selected walking speed. The database includes participants of varying ages, from 11 to 64 years. For each participant, up to eight gait analysis sessions were recorded, with each session comprising at least eight gait trials. The database provides unprocessed (raw) and processed (ready-to-use) data, including three-dimensional GRF and two-dimensional COP signals during the stance phase. These data records offer new possibilities for future studies on human gait, e.g., the application as a reference set for the analysis of pathological gait patterns, or for automatic classification using machine learning. In the future, the database will be expanded continuously to obtain an even larger and well-balanced database with respect to age, sex, and other gait-specific factors.

USING FAST FOURIER TRANSFORM AND POLYNOMIAL FITTING ON DORSAL FOOT KINEMATICS DATA TO IDENTIFY SIMULATED ANKLE SPRAIN MOTIONS FROM COMMON SPORTING MOTIONS

Ankle sprain is very common in sports, and a commonly suggested etiology is the delayed peroneal muscle reaction time. Recent studies showed the successful attempts to deliver electrical stimulation to the peroneal muscles externally to initiate contraction before it could react, however, the success relies on a workable method to detect ankle sprain injury in time. This study presented a fast Fourier transform and polynomial fitting method with dorsal foot kinematics data for quick ankle sprain detection. Five males performed 100 simulated ankle sprain and 250 common sporting motion trials. Eight gyrometers recorded the three-dimensional angular velocities at 500[Formula: see text]Hz. Data were trimmed with a 0.11[Formula: see text]s window size, the suggested duration of preinjury phase in ankle sprain, and were transformed from time to frequency domain by fast Fourier transform and fitted with a fifth-order polynomial. First-order coefficients from polynomial fitting on frequency space were obtained. The method achieved 97.0% sensitivity and 91.4% specificity in identifying simulated sprains, vertical jump–landing, cutting, stepping-down, running, and walking motions, with vertical jump–landing excluded due to its relatively low specificity (67.3%). The method can be used to detect ankle sprain in sports with mainly floor movements and minimal vertical jump–landing motion.

Walking stability in patients with benign paroxysmal positional vertigo: an objective assessment using wearable accelerometers and machine learning

BACKGROUND

Benign paroxysmal positional vertigo (BPPV) is one of the most common peripheral vestibular disorders leading to balance difficulties and increased fall risks. This study aims to investigate the walking stability of BPPV patients in clinical settings and propose a machine-learning-based classification method for determining the severity of gait disturbances of BPPV.

METHODS

Twenty-seven BPPV outpatients and twenty-seven healthy subjects completed level walking trials at self-preferred speed in clinical settings while wearing two accelerometers on the head and lower trunk, respectively. Temporo-spatial variables and six walking stability related variables [root mean square (RMS), harmonic ratio (HR), gait variability, step/stride regularity, and gait symmetry] derived from the acceleration signals were analyzed. A support vector machine model (SVM) based on the gait variables of BPPV patients were developed to differentiate patients from healthy controls and classify the handicapping effects of dizziness imposed by BPPV.

RESULTS

The results showed that BPPV patients employed a conservative gait and significantly reduced walking stability compared to the healthy controls. Significant different mediolateral HR at the lower trunk and anteroposterior step regularity at the head were found in BPPV patients among mild, moderate, and severe DHI (dizziness handicap inventory) subgroups. SVM classification achieved promising accuracies with area under the curve (AUC) of 0.78, 0.83, 0.85 and 0.96 respectively for differentiating patients from healthy controls and classifying the three stages of DHI subgroups. Study results suggest that the proposed gait analysis that is based on the coupling of wearable accelerometers and machine learning provides an objective approach for assessing gait disturbances and handicapping effects of dizziness imposed by BPPV.

Flash glucose monitoring data analysed by detrended fluctuation function on beta-cell function and diabetes classification

AIM

We aimed to use data-driven glucose pattern analysis to unveil the correlation between the metrics reflecting glucose fluctuation and beta-cell function, and to identify the possible role of this metric in diabetes classification.

MATERIALS AND METHODS

In total, 78 participants with type 1 diabetes and 59 with type 2 diabetes were enrolled in this study. All participants wore a flash glucose monitoring system, and glucose data were collected. A detrended fluctuation function (DFF) was utilized to extract glucose fluctuation information from flash glucose monitoring data and a DFF-based glucose fluctuation metric was proposed.

RESULTS

For the entire study population, a significant negative correlation between the DFF-based glucose fluctuation metric and fasting C-peptide was observed (r = -0.667; P <.001), which was larger than the correlation coefficient between the fasting C-peptide and mean amplitude of plasma glucose excursions (r = -0.639; P < .001), standard deviation (r = -0.649; P <.001), mean blood glucose (r = -0.519; P < .001) and time in range (r = 0.593; P < .001). As glucose data analysed by DFF revealed a clear bimodal distribution among the total participants, we randomly assigned the 137 participants into discovery cohorts (n = 100) and validation cohorts (n = 37) for 10 times to evaluate the consistency and effectiveness of the proposed metric for diabetes classification. The confidence interval for area under the curve according to the receiver operating characteristic analysis in the 10 discovery cohorts achieved (0.846, 0.868) and that for the 10 validation cohorts was (0.799, 0.862). In addition, the confidence intervals for sensitivity and specificity in the discovery cohorts were (75.5%, 83.0%), (81.3%, 88.5%) and (71.8%, 88.3%), (76.5%, 90.3%) in the validation cohorts, indicating the potential capacity of DFF in distinguishing type 1 and type 2 diabetes.

CONCLUSIONS

Our study first proposed the possible role of data-driven analysis acquired glucose metric in predicting beta-cell function and diabetes classification, and a large-scale, multicentre study will be needed in the future.

GaiTRec, a large-scale ground reaction force dataset of healthy and impaired gait

The quantification of ground reaction forces (GRF) is a standard tool for clinicians to quantify and analyze human locomotion. Such recordings produce a vast amount of complex data and variables which are difficult to comprehend. This makes data interpretation challenging. Machine learning approaches seem to be promising tools to support clinicians in identifying and categorizing specific gait patterns. However, the quality of such approaches strongly depends on the amount of available annotated data to train the underlying models. Therefore, we present GAITREC, a comprehensive and completely annotated large-scale dataset containing bi-lateral GRF walking trials of 2,084 patients with various musculoskeletal impairments and data from 211 healthy controls. The dataset comprises data of patients after joint replacement, fractures, ligament ruptures, and related disorders at the hip, knee, ankle or calcaneus during their entire stay(s) at a rehabilitation center. The data sum up to a total of 75,732 bi-lateral walking trials and enable researchers to classify gait patterns at a large-scale as well as to analyze the entire recovery process of patients.

Systematic Comparison of the Influence of Different Data Preprocessing Methods on the Performance of Gait Classifications Using Machine Learning

Human movements are characterized by highly non-linear and multi-dimensional interactions within the motor system. Therefore, the future of human movement analysis requires procedures that enhance the classification of movement patterns into relevant groups and support practitioners in their decisions. In this regard, the use of data-driven techniques seems to be particularly suitable to generate classification models. Recently, an increasing emphasis on machine-learning applications has led to a significant contribution, e.g., in increasing the classification performance. In order to ensure the generalizability of the machine-learning models, different data preprocessing steps are usually carried out to process the measured raw data before the classifications. In the past, various methods have been used for each of these preprocessing steps. However, there are hardly any standard procedures or rather systematic comparisons of these different methods and their impact on the classification performance. Therefore, the aim of this analysis is to compare different combinations of commonly applied data preprocessing steps and test their effects on the classification performance of gait patterns. A publicly available dataset on intra-individual changes of gait patterns was used for this analysis. Forty-two healthy participants performed 6 sessions of 15 gait trials for 1 day. For each trial, two force plates recorded the three-dimensional ground reaction forces (GRFs). The data was preprocessed with the following steps: GRF filtering, time derivative, time normalization, data reduction, weight normalization and data scaling. Subsequently, combinations of all methods from each preprocessing step were analyzed by comparing their prediction performance in a six-session classification using Support Vector Machines, Random Forest Classifiers, Multi-Layer Perceptrons, and Convolutional Neural Networks. The results indicate that filtering GRF data and a supervised data reduction (e.g., using Principal Components Analysis) lead to increased prediction performance of the machine-learning classifiers. Interestingly, the weight normalization and the number of data points (above a certain minimum) in the time normalization does not have a substantial effect. In conclusion, the present results provide first domain-specific recommendations for commonly applied data preprocessing methods and might help to build more comparable and more robust classification models based on machine learning that are suitable for a practical application.

Multiple Classification of Gait Using Time-Frequency Representations and Deep Convolutional Neural Networks

Human gait has served as a useful barometer of health. Existing studies for automatic categorization of gait have been limited to a binary classification of pathological and non-pathological gait and provided low accuracy in multi-classification. This study aimed to propose a novel approach that can multi-classify gait with no visually discernible difference in characteristics. Sixty-nine participants without gait disturbance were recruited. Twenty-nine of the participants were semi-professional athletes, and 19 were ordinary people. The remaining 21 participants were people with subtle foot deformities. The 3-axis acceleration and the 3-axis angular velocity signals were measured using the inertial measurement units attached to the foot, shank, thigh, and posterior pelvis while walking. The gait spectrograms were acquired by applying time-frequency analyses to the lower body movement signals measured in one stride and used to train the deep convolutional neural network-based classifiers. Four-fold cross-validation was applied to 80% of the total samples and the remaining 20% were used as test data. The foot, shank, and thigh spectrograms enabled complete classification of the three groups even without requiring a sophisticated process for feature engineering. This is the first study that employed the spectrographic approach in gait classification and achieved reliable multi-classification of gait without observable differences in characteristics using the deep convolutional neural networks.

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.

Estimation and validation of temporal gait features using a markerless 2D video system

BACKGROUND AND OBJECTIVE

Estimation of temporal gait features, such as stance time, swing time and gait cycle time, can be used for clinical evaluations of various patient groups having gait pathologies, such as Parkinson's diseases, neuropathy, hemiplegia and diplegia. Most clinical laboratories employ an optoelectronic motion capture system to acquire such features. However, the operation of these systems requires specially trained operators, a controlled environment and attaching reflective markers to the patient's body. To allow the estimation of the same features in a daily life setting, this paper presents a novel vision based system whose operation does not require the presence of skilled technicians or markers and uses a single 2D camera.

METHOD

The proposed system takes as input a 2D video, computes the silhouettes of the walking person, and then estimates key biomedical gait indicators, such as the initial foot contact with the ground and the toe off instants, from which several other temporal gait features can be derived.

RESULTS

The proposed system is tested on two datasets: (i) a public gait dataset made available by CASIA, which contains 20 users, with 4 sequences per user; and (ii) a dataset acquired simultaneously by a marker-based optoelectronic motion capture system and a simple 2D video camera, containing 10 users, with 5 sequences per user. For the CASIA gait dataset A the relevant temporal biomedical gait indicators were manually annotated, and the proposed automated video analysis system achieved an accuracy of 99% on their identification. It was able to obtain accurate estimations even on segmented silhouettes where, the state-of-the-art markerless 2D video based systems fail. For the second database, the temporal features obtained by the proposed system achieved an average intra-class correlation coefficient of 0.86, when compared to the ``gold standard" optoelectronic motion capture system.

CONCLUSIONS

The proposed markerless 2D video based system can be used to evaluate patients' gait without requiring the usage of complex laboratory settings and without the need for physical attachment of sensors/markers to the patients. The good accuracy of the results obtained suggests that the proposed system can be used as an alternative to the optoelectronic motion capture system in non-laboratory environments, which can be enable more regular clinical evaluations.

Falls Risk Classification of Older Adults Using Deep Neural Networks and Transfer Learning

Prior research in falls risk classification using inertial sensors has relied on the use of engineered features, which has resulted in a feature space containing hundreds of features that are likely redundant and possibly irrelevant. In this paper, we propose using fully convolutional neural networks (FCNNs) to classify older adults at low or high risk of falling using inertial sensor data collected from a smartphone. Due to the limited nature of older adult inertial gait datasets, we first pre-train the FCNN models using a publicly available dataset for pedestrian activity recognition. Then via transfer learning, we train the network for falls risk classification. We show that via transfer learning, our falls risk classifier obtains an area under the receiver operating characteristic curve of 93.3%, which is 10.6% higher than the equivalent model trained without the use of transfer learning. Additionally, we show that our method outperforms other standard machine learning classifiers trained on features developed in prior research.

Kinematic and Kinetic Patterns Related to Free-Walking in Parkinson's Disease

The aim of this study is to compare the properties of free-walking at a natural pace between mild Parkinson’s disease (PD) patients during the ON-clinical status and two control groups. In-shoe pressure-sensitive insoles were used to quantify the temporal and force characteristics of a 5-min free-walking in 11 PD patients, in 16 young healthy controls, and in 12 age-matched healthy controls. Inferential statistics analyses were performed on the kinematic and kinetic parameters to compare groups’ performances, whereas feature selection analyses and automatic classification were used to identify the signature of parkinsonian gait and to assess the performance of group classification, respectively. Compared to healthy subjects, the PD patients’ gait pattern presented significant differences in kinematic parameters associated with bilateral coordination but not in kinetics. Specifically, patients showed an increased variability in double support time, greater gait asymmetry and phase deviation, and also poorer phase coordination. Feature selection analyses based on the ReliefF algorithm on the differential parameters in PD patients revealed an effect of the clinical status, especially true in double support time variability and gait asymmetry. Automatic classification of PD patients, young and senior subjects confirmed that kinematic predictors produced a slightly better classification performance than kinetic predictors. Overall, classification accuracy of groups with a linear discriminant model which included the whole set of features (i.e., demographics and parameters extracted from the sensors) was 64.1%.