Integration of Task-Based Exoskeleton with an Assist-as-Needed Algorithm for Patient-Centered Elbow Rehabilitation

Area of Science:

• Biomedical Engineering
• Rehabilitation Robotics
• Machine Learning in Healthcare

Background:

• Elbow rehabilitation requires personalized and adaptive support.
• Existing robotic/exoskeleton devices need improved control strategies for patient engagement.
• Quantifying patient effort and progress in real-time is crucial for effective therapy.

Purpose of the Study:

• To develop and validate an Assist-as-Needed (AAN) algorithm for a bio-inspired elbow rehabilitation exoskeleton.
• To personalize the rehabilitation assistance based on individual patient capabilities.
• To provide real-time, motivating feedback to patients undergoing therapy.

Main Methods:

• Implementation of an Assist-as-Needed (AAN) algorithm using Force Sensitive Resistor (FSR) sensors.
• Integration of machine learning for personalized patient assistance.
• Utilizing Electromyography (EMG) signals and range of motion data for progress monitoring and feedback.
• Testing the system on participants with Spinal Cord Injury and Duchenne Muscular Dystrophy.

Main Results:

• The AAN algorithm achieved a high accuracy of 91.22% in controlling the exoskeleton during rehabilitation exercises.
• The system successfully provided real-time, visual feedback on elbow range of motion and patient effort.
• Quantification of disability levels was achieved by combining range of motion and FSR data.

Conclusions:

• The developed AAN algorithm offers effective, personalized rehabilitative support for elbow exoskeleton devices.
• Real-time feedback, combining biomechanical and sensor data, enhances patient motivation and therapy adherence.
• This approach represents a significant advancement in assistive technology for neurological and neuromuscular rehabilitation.