Real-Time Sensing of Upper Extremity Movement Diversity Using Kurtosis Implemented on a Smartwatch

Area of Science:

• Biomedical Engineering
• Rehabilitation Science
• Wearable Technology

Background:

• Traditional wearable sensors measure movement quantity (e.g., steps).
• Quantifying movement quality, such as diversity and complexity, is crucial for neurologic rehabilitation.
• Kurtosis, a statistical measure of distribution peakedness, can quantify forearm postural diversity in individuals with upper extremity impairment.

Purpose of the Study:

• To determine if different movement activities produce distinct kurtosis values.
• To develop a computationally efficient algorithm for real-time kurtosis calculation.
• To validate the real-time kurtosis algorithm on a commercial smartwatch for wearable rehabilitation applications.

Main Methods:

• Evaluated kurtosis of forearm postural diversity across 12 activities in 7 unimpaired individuals.
• Developed and implemented a novel recursive algorithm for real-time kurtosis computation.
• Validated the algorithm's accuracy on a smartwatch using a robotic simulator.

Main Results:

• Different activities resulted in a wide range of kurtosis values, indicating varied movement diversity.
• The novel algorithm reduced kurtosis computation time by a factor of 200.
• The smartwatch-based algorithm accurately calculated kurtosis during simulated activities.

Conclusions:

• Movement diversity, quantified by kurtosis, varies significantly across different activities.
• A computationally efficient algorithm enables real-time kurtosis calculation on wearable devices.
• This work is a key step towards developing wearable systems for real-time feedback on movement quality in rehabilitation.