Deep Learning for Accelerometric Data Assessment and Ataxic Gait Monitoring

Analysis of static plantar pressure data with capsule networks: Diagnosing ataxia in MS patients with a deep learning-based approach

In this study, it was aimed to detect ataxia in patients with Multiple Sclerosis (MS) by utilizing static plantar pressure data and capsule networks (CapsNet), one of the deep learning (DL) architectures. CapsNet is also equipped with a robust dynamic routing mechanism that determines the output of the next capsule. MS is a chronic nervous system disease that shows its effect in the central nervous system and manifests itself with attacks. One of the most common and challenging symptoms of MS is known as ataxia. Ataxia causes loss of control of limb muscle tone or gait disorders, leading to loss of balance and coordination. The diagnosis of ataxia in MS is applied employing the standard Expanded Disability Status Scale (EDSS) score. However, due to reasons such as physician misconception, diagnosis differences among physicians, and incorrect patient information, more unbiased solutions are required for the diagnosis. The results included Sensitivity at 96.34 % ± 1.71, Specificity at 98.11 % ± 2.04, Precision at 98.08 % ± 2.16, and Accuracy at 97.13 % ± 0.33. The main motivation of the study is to show that these deep learning methods can successfully detect ataxia in MS patients using static plantar pressure data. The high-performance measurements of sensitivity, specificity, precision and accuracy emphasize that the proposed system can be an effective tool in clinical practice. In addition, it was concluded that the proposed autonomous system would be a support mechanism to assist the physician in the detection of ataxia in patients with MS.

Wearable Sensor Technologies to Assess Motor Functions in People With Multiple Sclerosis: Systematic Scoping Review and Perspective

BACKGROUND

Wearable sensor technologies have the potential to improve monitoring in people with multiple sclerosis (MS) and inform timely disease management decisions. Evidence of the utility of wearable sensor technologies in people with MS is accumulating but is generally limited to specific subgroups of patients, clinical or laboratory settings, and functional domains.

OBJECTIVE

This review aims to provide a comprehensive overview of all studies that have used wearable sensors to assess, monitor, and quantify motor function in people with MS during daily activities or in a controlled laboratory setting and to shed light on the technological advances over the past decades.

METHODS

We systematically reviewed studies on wearable sensors to assess the motor performance of people with MS. We scanned PubMed, Scopus, Embase, and Web of Science databases until December 31, 2022, considering search terms "multiple sclerosis" and those associated with wearable technologies and included all studies assessing motor functions. The types of results from relevant studies were systematically mapped into 9 predefined categories (association with clinical scores or other measures; test-retest reliability; group differences, 3 types; responsiveness to change or intervention; and acceptability to study participants), and the reporting quality was determined through 9 questions. We followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) reporting guidelines.

RESULTS

Of the 1251 identified publications, 308 were included: 176 (57.1%) in a real-world context, 107 (34.7%) in a laboratory context, and 25 (8.1%) in a mixed context. Most publications studied physical activity (196/308, 63.6%), followed by gait (81/308, 26.3%), dexterity or tremor (38/308, 12.3%), and balance (34/308, 11%). In the laboratory setting, outcome measures included (in addition to clinical severity scores) 2- and 6-minute walking tests, timed 25-foot walking test, timed up and go, stair climbing, balance tests, and finger-to-nose test, among others. The most popular anatomical landmarks for wearable placement were the waist, wrist, and lower back. Triaxial accelerometers were most commonly used (229/308, 74.4%). A surge in the number of sensors embedded in smartphones and smartwatches has been observed. Overall, the reporting quality was good.

CONCLUSIONS

Continuous monitoring with wearable sensors could optimize the management of people with MS, but some hurdles still exist to full clinical adoption of digital monitoring. Despite a possible publication bias and vast heterogeneity in the outcomes reported, our review provides an overview of the current literature on wearable sensor technologies used for people with MS and highlights shortcomings, such as the lack of harmonization, transparency in reporting methods and results, and limited data availability for the research community. These limitations need to be addressed for the growing implementation of wearable sensor technologies in clinical routine and clinical trials, which is of utmost importance for further progress in clinical research and daily management of people with MS.

TRIAL REGISTRATION

PROSPERO CRD42021243249; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=243249.

Automatic ML-based vestibular gait classification: examining the effects of IMU placement and gait task selection

BACKGROUND

Vestibular deficits can impair an individual's ability to maintain postural and/or gaze stability. Characterizing gait abnormalities among individuals affected by vestibular deficits could help identify patients at high risk of falling and inform rehabilitation programs. Commonly used gait assessment tools rely on simple measures such as timing and visual observations of path deviations by clinicians. These simple measures may not capture subtle changes in gait kinematics. Therefore, we investigated the use of wearable inertial measurement units (IMUs) and machine learning (ML) approaches to automatically discriminate between gait patterns of individuals with vestibular deficits and age-matched controls. The goal of this study was to examine the effects of IMU placement and gait task selection on the performance of automatic vestibular gait classifiers.

METHODS

Thirty study participants (15 with vestibular deficits and 15 age-matched controls) participated in a single-session gait study during which they performed seven gait tasks while donning a full-body set of IMUs. Classification performance was reported in terms of area under the receiver operating characteristic curve (AUROC) scores for Random Forest models trained on data from each IMU placement for each gait task.

RESULTS

Several models were able to classify vestibular gait better than random (AUROC > 0.5), but their performance varied according to IMU placement and gait task selection. Results indicated that a single IMU placed on the left arm when walking with eyes closed resulted in the highest AUROC score for a single IMU (AUROC = 0.88 [0.84, 0.89]). Feature permutation results indicated that participants with vestibular deficits reduced their arm swing compared to age-matched controls while they walked with eyes closed.

CONCLUSIONS

These findings highlighted differences in upper extremity kinematics during walking with eyes closed that were characteristic of vestibular deficits and showed evidence of the discriminative ability of IMU-based automated screening for vestibular deficits. Further research should explore the mechanisms driving arm swing differences in the vestibular population.

The Effect of Face Masks on Physiological Data and the Classification of Rehabilitation Walking

Gait analysis and the assessment of rehabilitation exercises are important processes that occur during fitness level monitoring and the treatment of neurological disorders. This paper presents the possibility of using oximetric, heart rate (HR), accelerometric, and global navigation satellite systems (GNSSs) to analyse signals recorded during uphill and downhill walking without and with a face mask to find its influence on physiological functions during selected walking patterns. The experimental dataset includes 86 signal segments acquired under different conditions. The proposed methodology is based on signal analysis in both the time and frequency domains. The results indicate that face mask use has a minimal effect on blood oxygen concentration and heart rate, with the average mean changes of these parameters being less than 2%. The support vector machine, a Bayesian method, the k -nearest neighbour method, and a two-layer neural network showed very good separation abilities and successfully classified different walking patterns only in the case when the effect of face mask wearing was not included in the classification process. Our methodology suggests that artificial intelligence and machine learning tools are efficient methods for the assessment of motion patterns in different motion conditions and that face masks have a negligible effect for short-duration experiments.

PC-LSTM: Ontology-based Long Short-Term Memory State Model for Data Incompleteness Prediction

Medical practices are engaged and motivated by new technologies and methods to enhance patient care as efficiently as possible. These new methods and technologies give way for medical practices and clinicians to have the insight, comprehension, and projections to develop better decisions and overall levels of care. In this paper, we propose a model, PatientCentered-LSTM (or PC-LSTM), using the states of the LSTM model to produce a novel, ontology-based state system for data incompleteness. The overall architecture and system design are based around utilizing the hidden and cell states of the LSTM model to produce a network of states for each of the corresponding hierarchies in an Electronic Health Record (EHR) system. The resulting methodology allows for an accurate and precise approach to predicting data incompleteness in electronic health records. Clinical relevance- The method presented uses the hierarchical nature of electronic health record systems to positively influence the analysis of its data completeness; thereby, increasing the possibility of improved healthcare outcomes.

User Authentication by Gait Data from Smartphone Sensors Using Hybrid Deep Learning Network

User authentication and verification by gait data based on smartphones’ inertial sensors has gradually attracted increasing attention due to their compact size, portability and affordability. However, the existing approaches often require users to walk on a specific road at a normal walking speed to improve recognition accuracy. In order to recognize gaits under unconstrained conditions on where and how users walk, we proposed a Hybrid Deep Learning Network (HDLN), which combined the advantages of a long short-term memory (LSTM) network and a convolutional neural network (CNN) to reliably extract discriminative features from complex smartphone inertial data. The convergence layer of HDLN was optimized through a spatial pyramid pooling and attention mechanism. The former ensured that the gait features were extracted from more dimensions, and the latter ensured that only important gait information was processed while ignoring unimportant data. Furthermore, we developed an APP that can achieve real-time gait recognition. The experimental results showed that HDLN achieved better performance improvements than CNN, LSTM, DeepConvLSTM and CNN+LSTM by 1.9%, 2.8%, 2.0% and 1.3%, respectively. Furthermore, the experimental results indicated our model’s high scalability and strong suitability in real application scenes.

Federated Deep Learning for the Diagnosis of Cerebellar Ataxia: Privacy Preservation and Auto-crafted Feature Extractor

Cerebellar ataxia (CA) is concerned with the incoordination of movement caused by cerebellar dysfunction. Movements of the eyes, speech, trunk, and limbs are affected. Conventional machine learning approaches utilizing centralised databases have been used to objectively diagnose and quantify the severity of CA. Although these approaches achieved high accuracy, large scale deployment will require large clinics and raises privacy concerns. In this study, we propose an image transformation-based approach to leverage the advantages of state-of-the-art deep learning with federated learning in diagnosing CA. We use motion capture sensors during the performance of a standard neurological balance test obtained from four geographically separated clinics. The recurrence plot, melspectrogram, and poincaré plot are three transformation techniques explored. Experimental results indicate that the recurrence plot yields the highest validation accuracy (86.69%) with MobileNetV2 model in diagnosing CA. The proposed scheme provides a practical solution with high diagnosis accuracy, removing the need for feature engineering and preserving data privacy for a large-scale deployment.

Detection of ataxia with hybrid convolutional neural network using static plantar pressure distribution model in patients with multiple sclerosis

OBJECTIVE

In this study, it is aimed to detect ataxia for Persons with Multiple Sclerosis (PwMS) through a deep learning-based approach using an image dataset containing static plantar pressure distribution. Here, an alternative and objective method will be proposed to assist physicians who diagnose PwMS in the early stages.

METHODS

A total of 406 static bipedal pressure distribution image data for 43 ataxic PwMS and 62 healthy individuals were used in the study. After preprocessing, these images were given as input to pre-trained deep learning models such as VGG16, VGG19, ResNet, DenseNet, MobileNet, and NasNetMobile. The data of each model is utilized to generate its feature vectors. Finally, feature vectors obtained from static pressure distribution images were classified by SVM (Support Vector Machine), K-NN (K-Nearest Neighbors), and ANN (Artificial Neural Network). In addition, a cross-validation method was used to examine the validity of the classifier.

RESULTS

The performance of the proposed models was evaluated with accuracy, sensitivity, specificity, and F1-measure criteria. The VGG19-SVM hybrid model showed the best performance with 95.12% acc, 94.91% sen, 95.31% spe, and 94.44% F1.

CONCLUSIONS

In this study, a specific and sensitive automatic test evaluation system was proposed for Ataxic syndromes using digital images to observe the motor skills of the subjects. Comparative results show that the proposed method can be applied in practice for ataxia that is clinically difficult to detect or not yet symptomatic. It can be defined using only static plantar pressure distribution in the early stage and it can be recommended as an assistant system to physicians in clinical practice.

Incremental deep learning for reflectivity data recognition in stomatology

The recognition of stomatological disorders and the classification of dental caries are important areas of biomedicine that can hugely benefit from machine learning tools for the construction of relevant mathematical models. This paper explores the possibility of using reflectivity data to distinguish between healthy tissues and caries by deep learning and multilayer convolutional neural networks. The experimental data set includes more than 700 observations recorded in the stomatology laboratory. For rigor, the results obtained from the deep learning systems are compared with those evaluated for selected sets of features estimated for each observation and classified by a decision tree, support vector machine (SVM), k-nearest neighbor, Bayesian methods, and two-layer neural networks. The classification accuracy obtained for the deep learning systems was 98.1% and 94.4% for data in the signal and spectral domains, respectively, in comparison with an accuracy of 97.2% and 87.2% evaluated by the SVM method. The proposed method conclusively demonstrates how the artificial intelligence and deep learning methodology can contribute to improved diagnosis of dental problem in stomatology.

Analysis of Gait Sub-Movements to Estimate Ataxia Severity using Ankle Inertial Data

Objective: Assessment of motor severity in cerebellar ataxia is critical for monitoring disease progression and evaluating the effectiveness of therapeutic interventions. Though wearable sensors have been used to monitor gait tasks in order to enable frequent assessment, existing solutions only estimate gait performance severity rather than comprehensive motor severity. In this study, we propose a new approach that analyzes sub-second movement profiles of the lower-limbs during gait to estimate overall motor severity in cerebellar ataxia. Methods: A total of 37 ataxia subjects and 12 healthy subjects performed a 5 m walk-and-turn task with two ankle-worn inertial sensors. Lower-limb movements were decomposed into one-dimensional sub-movements, namely movement elements. Supervised regression models trained on data features of movement elements estimated the Brief Ataxia Rating Scale (BARS) and its sub-scores evaluated by clinicians. The proposed models were also compared to models trained on widely-accepted spatiotemporal gait features. Results: Estimated total BARS showed strong agreement with clinician-evaluated scores with r² = 0.72 and a root mean square error of 2.6 BARS points. Movement element-based models significantly outperformed conventional, spatiotemporal gait feature-based models. Conclusion: The proposed algorithm accurately assessed overall motor severity in cerebellar ataxia using inertial data collected from bilaterally-placed ankle sensors during a simple walk-and-turn task. Significance: Our work could support fine-grained monitoring of disease progression and patients' responses to medical/clinical interventions.

Detection of ataxia in low disability MS patients by hybrid convolutional neural networks based on images of plantar pressure distribution

BACKGROUND

This study aimed to detect ataxia in patients with multiple sclerosis (PwMS) with a deep learning-based approach based on images showing plantar pressure distribution of the patients. The secondary aim of the study was to investigate an alternative and objective method in the early diagnosis of ataxia in these patients.

METHODS

A total of 105 images showing plantar pressure distribution of 43 ataxic PwMS and 62 healthy individuals were analyzed. The images were resized for the models including VGG16, VGG19, ResNet, DenseNet, MobileNet, NasNetMobile, and NasNetLarge. Feature vectors were extracted from the resized images and then classified using Support Vector Machines (SVM), K-Nearest Neighbors (K-NN), and Artificial Neural Network (ANN). A 10-fold cross-validation was applied to increase the validity of the classifiers.

RESULTS

The VGG19-SVM hybrid model showed the highest accuracy, sensitivity, and specificity values (89.23%, 89.65%, and 88.88%, respectively).

CONCLUSION

The proposed method provided an automatic decision support system for detecting ataxia based on images showing plantar pressure distribution in patients with PwMS. The performance of the proposed method indicated that this method can be applied in clinical practice to establish a rapid diagnosis of ataxia that is asymptomatic or difficult to detect clinically and that it can be recommended as a useful aid for the physician in clinical practice.

Supporting the Assessment of Hereditary Transthyretin Amyloidosis Patients Based On 3-D Gait Analysis and Machine Learning

Hereditary Transthyretin Amyloidosis (vATTR-V30M) is a rare and highly incapacitating sensorimotor neuropathy caused by an inherited mutation (Val30Met), which typically affects gait, among other symptoms. In this context, we investigated the possibility of using machine learning (ML) techniques to build a model(s) that can be used to support the detection of the Val30Met mutation (possibility of developing the disease), as well as symptom onset detection for the disease, given the gait characteristics of a person. These characteristics correspond to 24 gait parameters computed from 3-D body data, provided by a Kinect v2 camera, acquired from a person while walking towards the camera. To build the model(s), different ML algorithms were explored: k-nearest neighbors, decision tree, random forest, support vector machines (SVM), and multilayer perceptron. For a dataset corresponding to 66 subjects (25 healthy controls, 14 asymptomatic mutation carriers, and 27 patients) and several gait cycles per subject, we were able to obtain a model that distinguishes between controls and vATTR-V30M mutation carriers (with or without symptoms) with a mean accuracy of 92% (SVM). We also obtained a model that distinguishes between asymptomatic and symptomatic carriers with a mean accuracy of 98% (SVM). These results are very relevant, since this is the first study that proposes a ML approach to support vATTR-V30M patient assessment based on gait, being a promising foundation for the development of a computer-aided diagnosis tool to help clinicians in the identification and follow-up of this disease. Furthermore, the proposed method may also be used for other neuropathies.