Random forest-based classsification and analysis of hemiplegia gait using low-cost depth cameras

Temporal Variability in Stride Kinematics during the Application of TENS: A Machine Learning Analysis

INTRODUCTION

The purpose of our report was to use a Random Forest classification approach to predict the association between transcutaneous electrical nerve stimulation (TENS) and walking kinematics at the stride level when middle-aged and older adults performed the 6-min test of walking endurance.

METHODS

Data from 41 participants (aged 64.6 ± 9.7 yr) acquired in two previously published studies were analyzed with a Random Forest algorithm that focused on upper and lower limb, lumbar, and trunk kinematics. The four most predictive kinematic features were identified and utilized in separate models to distinguish between three walking conditions: burst TENS, continuous TENS, and control. SHAP analysis and linear mixed models were used to characterize the differences among these conditions.

RESULTS

Modulation of four key kinematic features-toe-out angle, toe-off angle, and lumbar range of motion (ROM) in coronal and sagittal planes-accurately predicted walking conditions for the burst (82% accuracy) and continuous (77% accuracy) TENS conditions compared with control. Linear mixed models detected a significant difference in lumbar sagittal ROM between the TENS conditions. SHAP analysis revealed that burst TENS was positively associated with greater lumbar coronal ROM, smaller toe-off angle, and less lumbar sagittal ROM. Conversely, continuous TENS was associated with less lumbar coronal ROM and greater lumbar sagittal ROM.

CONCLUSIONS

Our approach identified four kinematic features at the stride level that could distinguish between the three walking conditions. These distinctions were not evident in average values across strides.

Machine learning algorithms integrate bulk and single-cell RNA data to unveil oxidative stress following intracerebral hemorrhage

BACKGROUND

Increased oxidative stress (OS) activity following intracerebral hemorrhage (ICH) had significantly impacting patient prognosis. Identifying optimal genes associated with OS could enhance the understanding of OS after ICH.

METHODS

We employed single-cell RNA sequencing (scRNA-seq) to investigate the heterogeneity of OS across various cellular tiers following ICH, aiming to acquire biological insights into ICH. We utilized AUCell, Ucell, singscore, ssgsea, and AddModuleScore algorithms, along with correlation analysis, to identify hub genes influencing high OS post-ICH. Furthermore, we employed four machine learning algorithms, eXtreme Gradient Boosting, Boruta, Random Forest, and Least Absolute Shrinkage and Selection Operator, to identify the optimal feature genes. To validate the accuracy of our analysis, we conducted validation in ICH animal experiments.

RESULTS

After analyzing the scRNA-seq dataset using various algorithms, we found that OS activity exhibited heterogeneity across different cellular layers following ICH, with particularly heightened activity observed in monocytes. Further integration of bulk data and machine learning algorithms revealed that ANXA2 and COTL1 were closely associated with high OS after ICH. Our animal experiments demonstrated an increase in OS expression post-ICH. Additionally, the protein expression of ANXA2 and COTL1 was significantly elevated and co-localized with microglia. Pearson correlation coefficient analysis revealed a significant correlation between ANXA2 and OS, indicating strong consistency (r = 0.84, p < 0.05). Similar results were observed for COTL1 and OS (r = 0.69, p < 0.05).

CONCLUSIONS

Following ICH, ANXA2 and COTL1 might penetrate the brain via monocytes, localize within microglia, and enhance OS activity. This might help us better understand OS after ICH.

Machine learning-based gait adaptation dysfunction identification using CMill-based gait data

Background

Combining machine learning (ML) with gait analysis is widely applicable for diagnosing abnormal gait patterns.

Objective

To analyze gait adaptability characteristics in stroke patients, develop ML models to identify individuals with GAD, and select optimal diagnostic models and key classification features.

Methods

This study was investigated with 30 stroke patients (mean age 42.69 years, 60% male) and 50 healthy adults (mean age 41.34 years, 58% male). Gait adaptability was assessed using a CMill treadmill on gait adaptation tasks: target stepping, slalom walking, obstacle avoidance, and speed adaptation. The preliminary analysis of variables in both groups was conducted using t-tests and Pearson correlation. Features were extracted from demographics, gait kinematics, and gait adaptability datasets. ML models based on Support Vector Machine, Decision Tree, Multi-layer Perceptron, K-Nearest Neighbors, and AdaCost algorithm were trained to classify individuals with and without GAD. Model performance was evaluated using accuracy (ACC), sensitivity (SEN), F1-score and the area under the receiver operating characteristic (ROC) curve (AUC).

Results

The stroke group showed a significantly decreased gait speed (p = 0.000) and step length (SL) (p = 0.000), while the asymmetry of SL (p = 0.000) and ST (p = 0.000) was higher compared to the healthy group. The gait adaptation tasks significantly decreased in slalom walking (p = 0.000), obstacle avoidance (p = 0.000), and speed adaptation (p = 0.000). Gait speed (p = 0.000) and obstacle avoidance (p = 0.000) were significantly correlated with global F-A score in stroke patients. The AdaCost demonstrated better classification performance with an ACC of 0.85, SEN of 0.80, F1-score of 0.77, and ROC-AUC of 0.75. Obstacle avoidance and gait speed were identified as critical features in this model.

Conclusion

Stroke patients walk slower with shorter SL and more asymmetry of SL and ST. Their gait adaptability was decreased, particularly in obstacle avoidance and speed adaptation. The faster gait speed and better obstacle avoidance were correlated with better functional mobility. The AdaCost identifies individuals with GAD and facilitates clinical decision-making. This advances the future development of user-friendly interfaces and computer-aided diagnosis systems.

Machine learning applied to gait analysis data in cerebral palsy and stroke: A systematic review

BACKGROUND

Among neurological pathologies, cerebral palsy and stroke are the main contributors to walking disorders. Machine learning methods have been proposed in the recent literature to analyze gait data from these patients. However, machine learning methods still fail to translate effectively into clinical applications. This systematic review addressed the gaps hindering the use of machine learning data analysis in the clinical assessment of cerebral palsy and stroke patients.

RESEARCH QUESTION

What are the main challenges in transferring proposed machine learning methods to clinical applications?

METHODS

PubMed, Web of Science, Scopus, and IEEE databases were searched for relevant publications on machine learning methods applied to gait analysis data from stroke and cerebral palsy patients until February the 23rd, 2023. Information related to the suitability, feasibility, and reliability of the proposed methods for their effective translation to clinical use was extracted, and quality was assessed based on a set of predefined questions.

RESULTS

From 4120 resulting references, 63 met the inclusion criteria. Thirty-one studies used supervised, and 32 used unsupervised machine learning methods. Artificial neural networks and k-means clustering were the most used methods in each category. The lack of rationale for features and algorithm selection, the use of unrepresentative datasets, and the lack of clinical interpretability of the clustering outputs were the main factors hindering the clinical reliability and applicability of these methods.

SIGNIFICANCE

The literature offers numerous machine learning methods for clustering gait data from cerebral palsy and stroke patients. However, the clinical significance of the proposed methods is still lacking, limiting their translation to real-world applications. The design of future studies must take into account clinical question, dataset significance, feature and model selection, and interpretability of the results, given their criticality for clinical translation.

The Identification of Elderly People with High Fall Risk Using Machine Learning Algorithms

Falling is an important public health issue, and predicting the fall risk can reduce the incidence of injury events in the elderly. However, most of the existing studies may have additional human and financial costs for community workers and doctors. Therefore, it is socially important to identify elderly people who are at high fall risk through a reasonable and cost-effective method. We evaluated the potential of multifractal, machine learning algorithms to identify the elderly at high fall risk. We developed a 42-point calibration model of the human body and recorded the three-dimensional coordinate datasets. The stability of the motion trajectory is calculated by the multifractal algorithm and used as an input dimension to compare the performance of the six classifiers. The results showed that the instability of the faller group was significantly greater than that of the no-faller group in the male and female cohorts (p < 0.005), and the Gradient Boosting Decision Tree classifier showed the best performance. The findings could help elderly people at high fall risk to identify individualized risk factors and facilitate tailored fall interventions.

Automatic Assessment of Functional Movement Screening Exercises with Deep Learning Architectures

(1) Background: The success of physiotherapy depends on the regular and correct unsupervised performance of movement exercises. A system that automatically evaluates these exercises could increase effectiveness and reduce risk of injury in home based therapy. Previous approaches in this area rarely rely on deep learning methods and do not yet fully use their potential. (2) Methods: Using a measurement system consisting of 17 inertial measurement units, a dataset of four Functional Movement Screening exercises is recorded. Exercise execution is evaluated by physiotherapists using the Functional Movement Screening criteria. This dataset is used to train a neural network that assigns the correct Functional Movement Screening score to an exercise repetition. We use an architecture consisting of convolutional, long-short-term memory and dense layers. Based on this framework, we apply various methods to optimize the performance of the network. For the optimization, we perform an extensive hyperparameter optimization. In addition, we are comparing different convolutional neural network structures that have been specifically adapted for use with inertial measurement data. To test the developed approach, it is trained on the data from different Functional Movement Screening exercises and the performance is compared on unknown data from known and unknown subjects. (3) Results: The evaluation shows that the presented approach is able to classify unknown repetitions correctly. However, the trained network is yet unable to achieve consistent performance on the data of previously unknown subjects. Additionally, it can be seen that the performance of the network differs depending on the exercise it is trained for. (4) Conclusions: The present work shows that the presented deep learning approach is capable of performing complex motion analytic tasks based on inertial measurement unit data. The observed performance degradation on the data of unknown subjects is comparable to publications of other research groups that relied on classical machine learning methods. However, the presented approach can rely on transfer learning methods, which allow to retrain the classifier by means of a few repetitions of an unknown subject. Transfer learning methods could also be used to compensate for performance differences between exercises.

Kinect-Based Assessment of Lower Limbs during Gait in Post-Stroke Hemiplegic Patients: A Narrative Review

The aim of this review was to present an overview of the state of the art in the use of the Microsoft Kinect camera to assess gait in post-stroke individuals through an analysis of the available literature. In recent years, several studies have explored the potentiality, accuracy, and effectiveness of this 3D optical sensor as an easy-to-use and non-invasive clinical measurement tool for the assessment of gait parameters in several pathologies. Focusing on stroke individuals, some of the available studies aimed to directly assess and characterize their gait patterns. In contrast, other studies focused on the validation of Kinect-based measurements with respect to a gold-standard reference (i.e., optoelectronic systems). However, the nonhomogeneous characteristics of the participants, of the measures, of the methodologies, and of the purposes of the studies make it difficult to adequately compare the results. This leads to uncertainties about the strengths and weaknesses of this technology in this pathological state. The final purpose of this narrative review was to describe and summarize the main features of the available works on gait in the post-stroke population, highlighting similarities and differences in the methodological approach and primary findings, thus facilitating comparisons of the studies as much as possible.

Prediction of human gait activities using wearable sensors

This paper aims to enhance the accuracy of human gait prediction using machine learning algorithms. Three classifiers are used in this paper: XGBoost, Random Forest, and SVM. A predefined dataset is used for feature extraction and classification. Gait prediction is determined during several locomotion activities: sitting (S), level walking (LW), ramp ascend (RA), ramp descend (RD), stair ascend (SA), stair descend (SD), and standing (ST). The results are gained for steady-state (SS) and overall (full) gait cycle. Two sets of sensors are used. The first set uses inertial measurement units only. The second set uses inertial measurement units, electromyography, and electro-goniometers. The comparison is based on prediction accuracy and prediction time. In addition, a comparison between the prediction times of XGBoost with CPU and GPU is introduced due to the easiness of using XGBoost with GPU. The results of this paper can help to choose a classifier for gait prediction that can obtain acceptable accuracy with fewer types of sensors.

Extraction and selection of gait recognition features using skeleton point detection and improved fuzzy decision

It is of great importance to effectively measure gait features and recognize the signature gait patterns for gait rehabilitation. In this work, we used a skeleton point detection to extract gait features and proposed an improved fuzzy decision to select the most significant features for classifying gait patterns. Thirteen gait recognition features were extracted from the obtained skeleton points data. Taking the extracted features as an input, our improved fuzzy similarity priority decision method has obtained important sequences of all features based on the relatively important scores. Then, the ranked features were delivered in different classifiers by a sequential forward selection strategy to select the optimal feature subset. There were significant differences between groups in each of the thirteen gait recognition features (p < 0.005), indicating that all extracted features are potential influence factors for classifying gait patterns. We also found that the highest classification accuracy of 100% for gait feature subsets included the stride frequency, maximum flexion angle of knee, and toe-out angle, on the all classifiers. The results suggest that the proposed approaches are very useful in searching for the optimal feature subset in present dataset.