Rehab-Net: Deep Learning Framework for Arm Movement Classification Using Wearable Sensors for Stroke Rehabilitation

Panwar, Madhuri; Biswas, Dwaipayan; Bajaj, Harsh P.; Jöbges, Michael; Turk, Ruth; Maharatna, Koushik; Acharyya, Amit

doi:10.1109/tbme.2019.2899927

Cited by 110 publications

(77 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alternative approach utilizes a machine learning algorithm to process complex sensor data and automatically extract a meaningful function (e.g. Neural Network model) that can classify the quality of motion [35,46,55]. However, no algorithm can completely replicate a therapist's assessment given diverse physical characteristics and functional abilities of patients.…”

Section: Motion Analysis Techniques For Rehabilitation Monitoring Andmentioning

confidence: 99%

“…Although Mansoor et al discusses the necessity of more investigation on challenges of accepting patient monitoring systems in clinic [3], a significant recent research has focused on improving the accuracy of monitoring and replicating clinician's decision making with a complex algorithm [35,46,49]. There is the lack of knowledge and evaluation [16] about therapist's experience on a decision support system for physical rehabilitation monitoring and assessment.…”

Section: Motion Analysis Techniques For Rehabilitation Monitoring Andmentioning

confidence: 99%

“…Despite the potential of clinical decision support systems, prior work has described the challenges of deploying these systems due to the lack of user-centered design consideration [9,27,65] and the opaqueness of machine learning algorithms [7,27,36,62]. However, significant recent research has focused on improving the accuracy of monitoring an exercise and replicating clinician's decision making with a complex algorithm [35,46,49]. Even if prior work using a complex algorithm makes a system more accurate, it is still difficult to derive a system that can perfectly replicate a therapist's decision making on rehabilitation assessment given diverse physical characteristics of patients [35].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Co-Design and Evaluation of an Intelligent Decision Support System for Stroke Rehabilitation Assessment

Lee

Siewiorek

Smailagic

et al. 2020

Proc. ACM Hum.-Comput. Interact.

View full text Add to dashboard Cite

Clinical decision support systems have the potential to improve work flows of experts in practice (e.g. therapist's evidence-based rehabilitation assessment). However, the adoption of these systems is challenging, and the gains of these systems have not fully demonstrated yet. In this paper, we identified the needs of therapists to assess patient's functional abilities (e.g. alternative perspectives with quantitative information on patient's exercise motions). As a result, we co-designed and developed an intelligent decision support system that automatically identifies salient features of assessment using reinforcement learning to assess the quality of motion and generate patient-specific analysis. We evaluated this system with seven therapists using the dataset from 15 patients performing three exercises. The results show that therapists have higher usage intent on our system than a traditional system without patient-specific analysis (< 0.05). While presenting richer information (< 0.10), our system significantly reduces therapists' effort on assessment (< 0.10) and improves their agreement on assessment from 0.66 to 0.71 F1-scores (< 0.01). This work discusses the importance of human centered design and development of a machine learning-based decision support system that presents contextually relevant information and salient explanations on its prediction for better adoption in practice. CCS Concepts: • Human-centered computing → Interactive systems and tools; User studies; • Applied computing → Health care information systems; • Computing methodologies → Reinforcement learning.

show abstract

Section: Motion Analysis Techniques For Rehabilitation Monitoring Andmentioning

confidence: 99%

Section: Motion Analysis Techniques For Rehabilitation Monitoring Andmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Co-Design and Evaluation of an Intelligent Decision Support System for Stroke Rehabilitation Assessment

Lee

Siewiorek

Smailagic

et al. 2020

Proc. ACM Hum.-Comput. Interact.

View full text Add to dashboard Cite

show abstract

“…Research presented by Guo, Gok & Sahin (2018) focused on implementation of the CNN model for prediction of the short-term dynamics of qualified forelimb movements based on neural signals in multiple animals. In Panwar et al (2019), the effective classification of three upper arm movements is presented. The ANN methods were used to assess rehabilitation based on Cao et al (2019), assess the progress of rehabilitation under the influence of a computer game (Bai, Song & Li, 2019) and analysis of the myoelectric signal during movement of upper limbs (Mukhopadhyay & Samui, 2020).…”

Section: Related Workmentioning

confidence: 99%

Convolutional neural network in upper limb functional motion analysis after stroke

Szczęsna

Błaszczyszyn

Kawala-Sterniuk

2020

PeerJ

View full text Add to dashboard Cite

In this work, implementation of Convolutional Neural Network (CNN) for the purpose of analysis of functional upper limb movement pattern was applied. The main aim of the study was to compare motion of selected activities of daily living of participants after stroke with the healthy ones (in similar age). The optical, marker-based motion capture system was applied for the purpose of data acquisition. There were some attempts made in order to find the existing differences in the motion pattern of the upper limb. For this purpose, the motion features of dominant and non-dominant upper limb of healthy participants were compared with motion features of paresis and non-paresis upper limbs of participants after stroke. On the basis of the newly collected data set, a new CNN application was presented to the classification of motion data in two different class label configurations. Analyzing individual segments of the upper body, it turned out that the arm was the most sensitive segment for capturing changes in the trajectory of the lifting movements of objects.

show abstract

“…Since, we have formulated this particular problem as a binary classification, therefore each task has two classes: class A and class B where class A is the 'specified task' and class B is the 'other'. The proposed model was formulated after exploring the hyper-parameter tuning [35] with different hyperparameters where 2 convolutional layers, 7 filters of size 5×1, stride rate of 1 in both convolutional layers and 2×1 maxpooling were chosen as optimum parameters for the given problem. This resultant in 213 neurons in the fully connected layer for LoCoMo-Net which is also illustrated in Fig 5. For training the model, categorical cross-entropy loss function, RMSprop optimizer with a default learning rate of 0.001 and the batch size of 15 are incorporated where maximum 50 epochs are used.…”

Section: Network Topology Of the Proposed Locomo-net Modelmentioning

confidence: 99%

Locomo-Net: A Low -Complex Deep Learning Framework for sEMG-Based Hand Movement Recognition for Prosthetic Control

Gautam

Panwar

Wankhede

et al. 2020

IEEE J. Transl. Eng. Health Med.

Self Cite

View full text Add to dashboard Cite

Background: The enhancement in the performance of the myoelectric pattern recognition techniques based on deep learning algorithm possess computationally expensive and exhibit extensive memory behavior. Therefore, in this paper we report a deep learning framework named 'Low-Complex Movement recognition-Net' (LoCoMo-Net) built with convolution neural network (CNN) for recognition of wrist and finger flexion movements; grasping and functional movements; and force pattern from single channel surface electromyography (sEMG) recording. The network consists of a two-stage pipeline: 1) input data compression; 2) data-driven weight sharing. Methods: The proposed framework was validated on two different datasets-our own dataset (DS1) and publicly available NinaPro dataset (DS2) for 16 movements and 50 movements respectively. Further, we have prototyped the proposed LoCoMo-Net on Virtex-7 Xilinx field-programmable gate array (FPGA) platform and validated for 15 movements from DS1 to demonstrate its feasibility for real-time execution. Results: The effectiveness of the proposed LoCoMo-Net was verified by a comparative analysis against the benchmarked models using the same datasets wherein our proposed model outperformed Twin-Support Vector Machine (SVM) and existing CNN based model by an average classification accuracy of 8.5 % and 16.0 % respectively. In addition, hardware complexity analysis is done to reveal the advantages of the two-stage pipeline where approximately 27 %, 49 %, 50 %, 23 %, and 43 % savings achieved in lookup tables (LUT's), registers, memory, power consumption and computational time respectively. Conclusion: The clinical significance of such sEMG based accurate and lowcomplex movement recognition system can be favorable for the potential improvement in quality of life of an amputated persons.

show abstract

Rehab-Net: Deep Learning Framework for Arm Movement Classification Using Wearable Sensors for Stroke Rehabilitation

Cited by 110 publications

References 49 publications

Co-Design and Evaluation of an Intelligent Decision Support System for Stroke Rehabilitation Assessment

Co-Design and Evaluation of an Intelligent Decision Support System for Stroke Rehabilitation Assessment

Convolutional neural network in upper limb functional motion analysis after stroke

Locomo-Net: A Low -Complex Deep Learning Framework for sEMG-Based Hand Movement Recognition for Prosthetic Control

Contact Info

Product

Resources

About