Exploratory development of human-machine interaction strategies for post-stroke upper-limb rehabilitation

Stroke and its related complications, place significant burdens on human society in the twenty-first century, and lead to substantial demands for upper limb rehabilitation. To fulfill the rehabilitation needs, human-machine interaction (HMI) technology strives continuously. Depends on the involvement of subject, HMI strategy can be classified as passive or active. Compare to passive modalities, active strategies are believed to be more effective in promoting neuroplasticity and motor recovery for post-stroke survivors in sub-acute and chronic phase. However, post-stroke survivors usually experience weak upper arms, limited range of motion (ROM) and involuntary excessive movement patterns. Distinguishing between complex subtle motion intentions and excessive involuntary movements in real-time remains a challenge in current research, which impedes the application of active HMI strategies in clinical practice.

An Up-limb Rehabilitation Device and Utility System (UarDus) is proposed along with 3 HMI strategies namely robot-in-charge, therapist-in-charge and patient-in-charge. Based on physiological structure of human upper-limb and scapulohumeral rhythm (SHR) of shoulder, a base exoskeleton with 14 degrees of freedoms (DoFs) is designed as foundation of the 3 strategies. Passive robot-in-charge and therapist-in-charge strategies provides fully-assisted rehabilitation options. The active patient-in-charge strategy incorporates data acquisition matrices and a new deep learning model, which is developed based on Convolutional Neural Network (CNN) and Transformer structure, aims to capture subtle motion intentions. Motors' current is monitored and the surge in the current is identified adopting Discrete Wavelet Transform (DWT) method for safety concerns.

Kinematically, the work space of the base exoskeleton is presented first. Utilizing motion capture technology, the glenohumeral joint (GH) centers of both human and exoskeleton exhibit well-matched motion curves, suggesting a comfortable dynamic wear experience. For robot-in-charge and therapist-in-charge strategy, the desired and measured angle-time curve present good correlation, with low phase difference, which serve the purpose of real-time control. Featuring the patient-in-charge strategy, Kernel Density Estimation (KDE) result suggesting reasonable sensor-machine-human synergy. Applying K-fold (K = 10) cross-validation method, the classification accuracy of the proposed model with outstanding response time achieves an average of 99.7% for the designated 15 actions, signifies its capability for subtle motion intention recognition in real-time. Additionally, signal surge is easily identified with DWT.

An upper-limb exoskeleton hardware device named UarDus is constructed, along with three HMI modalities, offering both passive and active rehabilitation approaches. The proposed system is validated through a proof-of-concept study on a subject who underwent a craniotomy for a hemorrhagic stroke, demonstrating the possibility for post-stroke individuals to engage in safe, personalized rehabilitation training in real-time, with a dynamically comfortable wear experience.