Detecting compensatory movements of stroke survivors using pressure distribution data and machine learning algorithms

Abstract

Background: Compensatory 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. Methods: Eight stroke survivors performed three types of reaching tasks (back-and-forth, side-to-side, and up-anddown 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. Results: Cross-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 F1score 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. Conclusions: Good 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.

Deep learning based behavioral profiling of rodent stroke recovery

“…Compensatory strategies (e.g. avoiding the use of the impaired limb or relying on the intact limb) are highly prevalent in rodents and in humans 34,35 . Although functional recovery is generally observed in a variety of tests, it is important to distinguish between compensatory responses and "true" recovery.…”

A novel time-based surface EMG measure for quantifying hypertonia in paretic arm muscles during daily activities after hemiparetic stroke

“…As such, they have been employed in many applications in rehabilitation, e.g. to recognize motor pattern or intent for robotic prosthetics control [29][30][31], to detect disease symptoms or adverse events [32][33][34], or to provide bio-feedback during training [35]. However, there are remaining gaps in utilizing these technologies to constantly monitor an important neural signature underlying post-stroke motor impairment such as hypertonia.…”

Development of a Compensation-aware Virtual Rehabilitation System for Upper Extremity Rehabilitation in Community-Dwelling Older Adults with Stroke

“…Recent studies start addressing the compensation in the design of VR-based rehabilitation through applying various motion tracking technology [16,17, 18] as well as the pressure sensors [19,20] to detect compensatory movements when stroke survivors perform the UE reaching tasks. For example, Foreman and Engsberg [17] utilized one commercial Kinect device in their VRShape system to measure and shape the trunk compensatory movement during repetitive upper limb practices of a reaching task.…”

“…For example, Cirstea and Levin [10] found a signi cant correlation between the increased trunk involvement with a reduction in active ranges of elbow extension and shoulder exion when stroke individuals were asked to perform the reaching task. The monitoring of stroke survivor's compensation during motor practice is especially important at the context of community where the close supervision of therapist is usually missed.Recent studies start addressing the compensation in the design of VR-based rehabilitation through applying various motion tracking technology [16,17, 18] as well as the pressure sensors [19,20] to detect compensatory movements when stroke survivors perform the UE reaching tasks. For example, Foreman and Engsberg [17] utilized one commercial Kinect device in their VRShape system to measure and shape the trunk compensatory movement during repetitive upper limb practices of a reaching task.…”

“…Besides the trunk compensation, the affected limb could play a role in compensation, for example, the affected upper arm would compensate the lower arm motion in elevating the upper limb [21]. Cai's studies [19,20] could further detect the shoulder elevation besides trunk compensations using the pressure mattress sit under the stroke survivors. Thus, to reduce the compensation in the UE motor practice, it is necessary to measure both the trunk and upper limb compensations.…”