Early recognition of upper limb motor tasks through accelerometers: real-time implementation of a DTW-based algorithm

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Human-Computer Interaction

Background:

• Accurate classification of human movements is crucial for developing effective rehabilitation and assistive technologies.
• Traditional Dynamic Time Warping (DTW) methods offer robust movement classification but lack real-time capabilities for interactive applications.

Purpose of the Study:

• To introduce and evaluate a novel real-time Dynamic Time Warping (DTW)-based classification scheme for human movements.
• To assess the accuracy and timeliness of the real-time DTW system compared to traditional DTW methods.
• To explore the potential of this technology in real-time interactive applications for rehabilitation and assistive purposes.

Main Methods:

• Nine healthy adults performed eight distinct movements from the Wolf Motor Function Test.
• A three-axis accelerometer sensor was worn on the inner forearm to capture movement data.
• The performance of the real-time DTW classification was evaluated against a traditional DTW approach, analyzing correct recognition rates and recognition timing.

Main Results:

• The real-time DTW classification scheme achieved a high correct recognition percentage of 91.5%.
• Adoption of a multidimensional scheme further improved accuracy to 96.5%.
• Over 60% of movements were correctly recognized before their completion, demonstrating significant real-time capability.

Conclusions:

• The developed real-time DTW system demonstrates comparable accuracy to traditional DTW methods.
• The ability to recognize movements before completion opens avenues for real-time interaction in assistive and rehabilitation technologies.
• This advancement facilitates the development of responsive systems that can engage with users during the execution of movements.