Smart IoT-Based Wearable with Sensing for Functional Rehabilitation

High-dosage lower-limb motor practice has been proven in many existing studies that could effectively help with patients’ functional recovery. Yet, evaluation of to which extent the patient adheres to the practice and the performances are not considerably satisfied and many challenges yet to be solved before adopted to clinical practice. Up to date, research that aims at utilizing assistive technology for functional rehabilitation monitoring, such as using 3D motion capture system and depth camera, focused on the performance evaluation using pre-recorded activities data. Few studies have researched on classifying the rehabilitation exercises and their effectiveness in real time in the clinical setting (such as the hospital). This work proposes a low-cost
unobstructive IoT-based wearable solution that could perform tracking, classification and quality assessment of patients' rehabilitation performances more effectively where the rehabilitation tasks are based on the American
Academy of Orthopedic Surgeons (AAOS). The study utilizes four thumb-size micro-controllers with an integrated six degree-of-freedom (DoF) motion sensor placed on the dorsum of foot and lower shin to capture the motion of
patients while performing rehabilitation tasks. Then, a fine-tuned machine learning model is trained to provide the assessment of a patient’s rehabilitation tasks based on the real-time motion data. The quality assessment of
the rehabilitation is visualized in a graphical form to assist clinicians to evaluate the training outcome and further update the training plans to fit the patient’s needs.