Automatic Detection of Compensatory Movement Patterns by a Pressure Distribution Mattress Using Machine Learning Methods: A Pilot Study

BackgroundCompensatory movements are commonly employed by stroke survivors during seated reaching and may have negative effects on their long-term recovery. Detecting compensation is useful for coaching the patient to reduce compensatory trunk movements and improving the motor function of the paretic arm. Sensor-based and camera-based systems have been developed to detect compensatory movements, but they still have some limitations, such as causing object obstructions, requiring complex setups and raising privacy concerns. To overcome these drawbacks, this paper proposes a compensatory movement detection system based on pressure distribution data and is unobtrusive, simple and practical. Machine learning algorithms were applied to classify compensatory movements automatically. Therefore, the purpose of this study was to develop and test a pressure distribution-based system for the automatic detection of compensation movements of stroke survivors using machine learning algorithms.MethodsEight stroke survivors performed three types of reaching tasks (back-and-forth, side-to-side, and up-and-down reaching tasks) with both the healthy side and the affected side. The pressure distribution data were recorded, and five features were extracted for classification. The k-nearest neighbor (k-NN) and support vector machine (SVM) algorithms were applied to detect and categorize the compensatory movements. The surface electromyography (sEMG) signals of nine trunk muscles were acquired to provide a detailed description and explanation of compensatory movements.ResultsCross-validation yielded high classification accuracies (F1-score>0.95) for both the k-NN and SVM classifiers in detecting compensation movements during all the reaching tasks. In detail, an excellent performance was achieved in discriminating between compensation and noncompensation (NC) movements, with an average F1-score of 0.993. For the multiclass classification of compensatory movement patterns, an average F1-score of 0.981 was achieved in recognizing the NC, trunk lean-forward (TLF), trunk rotation (TR) and shoulder elevation (SE) movements.ConclusionsGood classification performance in detecting and categorizing compensatory movements validated the feasibility of the proposed pressure distribution-based system. Reliable classification accuracy achieved by the machine learning algorithms indicated the potential to monitor compensation movements automatically by using the pressure distribution-based system when stroke survivors perform seated reaching tasks.

The Effects of Constraint-Induced Therapy on Kinematic Outcomes and Compensatory Movement Patterns: An Exploratory Study

The aims of this study were to distinguish between behavioral compensation and behavioral recovery and to determine the role of stroke lesions and the optimal timing of rehabilitation in true recovery. Single pellet reaching test has been performed to analyze both quantitative and qualitative measures of forelimb function in a stroke animal model with lesions in the motor cortex, somatosensory cortex, or sensorimotor cortex. The four gestures of compensatory movement patterns that comprised a reach were head lift, limb withdrawal, pellet chasing, and phantom grasp. Functional recovery improved in all the stroke groups after rehabilitation ( P < 0.001). However, the compensatory movement patterns of the motor cortex and somatosensory cortex stroke groups initially increased and subsequently decreased ( P = 0.0054), whereas those of the sensorimotor cortex stroke group increased and persisted ( P = 0.0063). In the sensorimotor cortex stroke group, compensatory movement patterns significantly decreased when training was initiated 5 and 14 days after stroke ( P = 0.0083, P = 0.0226, respectively), while they increased and persisted when training was initiated 1 day after stroke. These findings suggest that true recovery by task-specific training after stroke depends, probably, on the lesion size and the timing of rehabilitation.

Kinematic redundancy of the human body provides abundant movement patterns to accomplish the same motor goals (motor equivalence). Compensatory movement patterns such as excessive trunk displacement in stroke subjects during reaching can be viewed as a consequence of the motor equivalent process to accomplish a task despite limited available ranges in some joints. However, despite compensations, the ability to adapt reaching performance when perturbations occur may still be limited when condition-specific changes of joint angles are required. We addressed this hypothesis in individuals with and without stroke for reaching a target placed beyond arm reach in standing while flexing the hips (free-hip condition). In randomly selected trials, hip flexion was unexpectedly blocked, forcing subjects to take a step (blocked-hip condition). In additional trials, subjects took an intentional step while reaching the target (intentional-step condition). In blocked-hip trials, healthy subjects maintained smooth and precise endpoint trajectories by adapting temporal and spatial interjoint coordination to neutralize the effect of the perturbation. However, the ability to produce motor equivalent solutions was reduced in subjects with stroke, evidenced by substantial overshoot errors in endpoint position, reduced movement smoothness and less adaptive elbow-shoulder interjoint coordination. Movement adaptability was more limited in stroke subjects who used more compensatory movements for unperturbed reaching. Results suggest that subjects with mild-to-moderate stroke only partially adapted arm joint movements to maintain reaching performance. Therapeutic efforts to enhance the ability of individuals with stroke to find a larger number of task-relevant motor solutions (adaptability) may improve upper limb recovery.

High-intensity, high-frequency exercise loads render muscle, ligament, and joint injuries among the most common ailments in training and competition. Affected athletes often experience persistent muscle weakness, compensatory movement patterns, and recurrent injuries. In recent years, the application of wearable devices in sports rehabilitation has mitigated these negative effects. Common types include surface electromyography (sEMG) and accelerometers. Compared to traditional exercise therapy, wearable technology integrates quantitative data, providing a more reliable understanding of the athlete’s injury and helping physicians develop more effective rehabilitation plans. However, when analyzing the effectiveness of these two wearable technologies, accelerometers and surface electromyography, several potential factors and limitations still need to be considered, such as the high cost, which may limit their widespread adoption and effectiveness in the field of sports therapy. Based on existing clinical and sports science research, this paper evaluates the effectiveness of sEMG and accelerometers in sports therapy, focusing on their benefits in movement quality assessment, training load management, and individualized rehabilitation plan development. It also briefly explores the potential of combining wearable data with artificial intelligence modeling for injury risk assessment and rehabilitation decision support.

In lower-limb pathomechanical patient populations, one of the greatest impediments to energy-efficient ambulation is the development of a genu recurvatum deformity. When this deviation occurs, the thigh and lower limb segments move posteriorly in direct opposition to the anterior advancement of the proximal body mass over the fixed distal base of support. Genu recurvatum usually is an acquired deformity secondary to changes of the distal skeletal joint alignments and compensatory movement patterns. The alignment of the distal base of support determines the load bearing and functional motion available to the proximal knee joint during ambulation and can serve either to promote or disrupt the required sagittal plane limb advancement This article describes three distinct pathomechanical profiles of posterior knee joint deviation in relation to the primary planes of motion. The distinction between an initial hyperextension moment and an acquired recurvatum deformity is presented in terms of predictability, prevention and correction of joint deformation. Gait training and orthotic design criteria are discussed as they relate to a comprehensive rehabilitation program to maximize both structural and functional outcomes for patients with lower-limb instabilities and deformities

ABSTRACTLeg amputation leads to irreversible limb loss and physical impairments, which requires local and systemic skeletal adaptations. The functional morphology of amputees in antiquity has not been systematically investigated. In this study, an individual with signs of lower limb amputation who lived 2200 years ago during the Warring States Period of the Eastern Zhou Dynasty was assessed using functional morphological approaches. The skeleton demonstrated functional morphological adaptations both locally and systematically. Bone thickness and enthesis changes of the amputated leg had signs of atrophy, indicating functional decline and prolonged disuse after the amputation. After amputation, the individual relied on his upper limbs for balance and support. Besides, the individual experienced considerable lumbar strain attributable to his postamputation lifestyle and compensatory movement patterns. The reason for amputation might be penal in a military setting. These findings reveal adaptive mechanisms of limb function and the challenges faced by amputees, offering insights into the consequences of such procedures.

Prediction of chronic kidney disease (CKD) has emerged as a useful technique for early detection of at-risk persons and the introduction of appropriate management strategies. Machine learning and data-driven methods have been used in predictive modeling to examine massive databases of patient demographics, medical histories, test findings, and genetic information. These cutting-edge methods allow for the profiling of high-risk patients and the tailoring of healthcare administration approaches. Patient outcomes, complication rates, and healthcare system efficiency may all benefit greatly from CKD screening and prediction. Responsible use of CKD prediction algorithms, however, requires resolving issues with data availability, integration, and ethics. The area of medicine has benefited greatly from the use of Machine Learning (ML) methods, which have played an increasingly central role in illness prediction. In this study, we use a strategy that makes use of ML methods to construct effective tools for predicting the development of CKD. Multiple ML models are trained, and their results are compared using a variety of criteria. We applied five ML methods such as logistic regression (LR), Decision tree (DT), random forest (RF), support vector machine (SVM), and k-nearest neighbor (KNN). The LR and KNN have the highest accuracy with 99%.

Industrial robots are widely used in critical tasks such as assembly, welding, and material handling as core components of modern manufacturing systems. For the reliable operation of these systems, early and accurate detection of execution failures is crucial. In this study, a comprehensive comparison of machine learning and deep learning methods is conducted for the classification of robot execution failures using data acquired from force-torque sensors. Three different feature engineering approaches are proposed. The first is a Baseline approach that includes 90 raw time-series features. The second is the Domain-6 approach, which consists of 6 basic statistical features per sensor (36 in total). The third is the Domain-12 approach, which comprises 12 comprehensive statistical features per sensor (72 in total). The domain features include the mean, standard deviation, minimum, maximum, range, slope, median, skewness, kurtosis, RMS, energy, and IQR. In total, ten classification algorithms are evaluated, including eight machine learning methods and two deep learning models: Support Vector Machines (SVM), Random Forest (RF), k-Nearest Neighbors (KNN), Artificial Neural Network (ANN), Naive Bayes (NB), Decision Trees (DT), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM-LGBM), as well as a One-Dimensional Convolutional Neural Network (CNN-1D) and Long Short-Term Memory (LSTM). For traditional machine learning algorithms, 5 × 5 nested cross-validation is used, whereas for deep learning models, 5-fold cross-validation with a 20% validation split is employed. To ensure statistical reliability, all experiments are repeated over 30 independent runs. The experimental results demonstrate that feature engineering has a decisive impact on classification performance. In addition, regardless of the feature set, the highest accuracy (93.85% ± 0.90) is achieved by the Naive Bayes classifier using the Baseline features. The Domain-12 feature set provides consistent improvements across many algorithms, with substantial performance gains. The results are reported using accuracy, precision, recall, and F1-score metrics and are supported by confusion matrices. Finally, permutation feature importance analysis indicates that the skewness features of the Fx and Fy sensors are the most critical variables for failure detection. Overall, these findings show that time-domain statistical features offer an effective approach for robot failure classification.

The quality of arm movements typically improves in the sub-acute phase of stroke affecting the upper extremity. Here, we used whole arm kinematic analysis during reaching movements to distinguish whether these improvements are due to true recovery or to compensation. Fifty-three participants with post-acute stroke performed ∼80 reaching movement tests during 4 weeks of training with the ArmeoSpring exoskeleton. All participants showed improvements in end-effector performance, as measured by movement smoothness. Four ArmeoSpring angles, shoulder horizontal (SH) rotation, shoulder elevation (SE), elbow rotation, and forearm rotation, were recorded and analyzed. We first characterized healthy joint coordination patterns by performing a sparse principal component analysis on these four joint velocities recorded during reaching tests performed by young control participants. We found that two dominant joint correlations [SH with elbow rotation and SE with forearm rotation] explained over 95% of variance of joint velocity data. We identified two clusters of stroke participants by comparing the evolution of these two correlations in all tests. In the “Recoverer” cluster (N = 19), both joint correlations converged toward the respective correlations for control participants. Thus, Recoverers relearned how to generate smooth end-effector movements while developing joint movement patterns similar to those of control participants. In the “Compensator” cluster (N = 34), at least one of the two joint correlations diverged from the corresponding correlation of control participants. Compensators relearned how to generate smooth end-effector movements by discovering various new compensatory movement patterns dissimilar to those of control participants. New compensatory patterns included atypical decoupling of the SE and forearm joints, and atypical coupling of the SH rotation and elbow joints. There was no difference in clinical impairment level between the two groups either at the onset or at the end of training as assessed with the Upper Extremity Fugl-Meyer scale. However, at the start of training, the Recoverers showed significantly faster improvements in end-effector movement smoothness than the Compensators. Our analysis can be used to inform neurorehabilitation clinicians on how to provide movement feedback during practice and suggest avenues for refining exoskeleton robot therapy to reduce compensatory patterns.

The huge resources need effectiveness and efficiency, it can be processed by machine learning. There have been many studies conducted using machine learning method and produced quite good performance in sentiment analysis. Some machine learning methods that are often used in general are Naive bayes (NB), K-nearest neighbor (KNN), Support vector machine (SVM), and Random forest methods. Mostly, KNN did not achieve better performance than other machine learning methods in sentiment analysis. In this study, the Polarity v2.0 from Cornell movie review dataset will be used to test KNN with Information gain features selection in order to achieve good performance. The purpose of this research are to find the optimum K for KNN and compare KNN with other methods. KNN with the help of Information gain feature selection becomes the best performance method with 96.8% accuracy compared to the NB, SVM, and Random forest while the optimum K is 3.

Objective: Movement impairments such as those due to stroke often result in the nervous system adopting atypical movements to compensate for movement deficits. Monitoring these compensatory patterns is critical for improving functional outcomes during rehabilitation. The purpose of this study was to test the feasibility and validity of a wearable sensor system for detecting compensatory trunk kinematics during activities of daily living. Approach: Participants with no history of neurological impairments performed reaching and manipulation tasks with their upper extremity, and their movements were recorded by a wearable sensor system and validated using a motion capture system. Compensatory movements of the trunk were induced using a brace that limited range of motion at the elbow. Main results: Our results showed that the elbow brace elicited compensatory movements of the trunk during reaching tasks but not manipulation tasks, and that a wearable sensor system with two sensors could reliably classify compensatory movements (~90% accuracy). Significance: These results show the potential of the wearable system to assess and monitor compensatory movements outside of a lab setting.

ABSTRACTIn the presented study, the Sentinel-2 vegetation indices (VIs) were evaluated in context of estimating defoliation of Scots pine stands in western Poland. Regression and classification models were built based on reference data from 50 field plots and Sentinel-2 satellite images from three acquisition dates. Three machine-learning (ML) methods were tested: k-nearest neighbors (kNN), random forest (RF), and support vector machines (SVM). Regression models predicted stands defoliation with moderate accuracy. R2 values for regression models amounted to 0.53, 0.57, 0.57 for kNN, RF and SVM, accordingly. Analogically, the following values of normalized root mean squared error were obtained: 12.2%, 11.9% and 11.6%. Overall accuracies for two-class classification models were 78%, 75%, 78% for kNN, RF and SVM methods. The Green Normalized Difference Vegetation Index and MERIS Terrestrial Chlorophyll Index VIs were found to be most robust defoliation predictors regardless of the ML method. We conclude that Sentinel-2 satellite images provide useful information about forest defoliation and may contribute to forest monitoring systems.

Artificial intelligence has been applied in wildfire science and management since the 1990s, with early applications including neural networks and expert systems. Since then, the field has rapidly progressed congruently with the wide adoption of machine learning (ML) methods in the environmental sciences. Here, we present a scoping review of ML applications in wildfire science and management. Our overall objective is to improve awareness of ML methods among wildfire researchers and managers, as well as illustrate the diverse and challenging range of problems in wildfire science available to ML data scientists. To that end, we first present an overview of popular ML approaches used in wildfire science to date and then review the use of ML in wildfire science as broadly categorized into six problem domains, including (i) fuels characterization, fire detection, and mapping; (ii) fire weather and climate change; (iii) fire occurrence, susceptibility, and risk; (iv) fire behavior prediction; (v) fire effects; and (vi) fire management. Furthermore, we discuss the advantages and limitations of various ML approaches relating to data size, computational requirements, generalizability, and interpretability, as well as identify opportunities for future advances in the science and management of wildfires within a data science context. In total, to the end of 2019, we identified 300 relevant publications in which the most frequently used ML methods across problem domains included random forests, MaxEnt, artificial neural networks, decision trees, support vector machines, and genetic algorithms. As such, there exists opportunities to apply more current ML methods — including deep learning and agent-based learning — in the wildfire sciences, especially in instances involving very large multivariate datasets. We must recognize, however, that despite the ability of ML models to learn on their own, expertise in wildfire science is necessary to ensure realistic modelling of fire processes across multiple scales, while the complexity of some ML methods such as deep learning requires a dedicated and sophisticated knowledge of their application. Finally, we stress that the wildfire research and management communities play an active role in providing relevant, high-quality, and freely available wildfire data for use by practitioners of ML methods.

Integrated phenology and climate in rice yields prediction using machine learning methods