Measurement of Functional Use in Upper Extremity Prosthetic Devices Using Wearable Sensors and Machine Learning.

Area of Science:

• Biomechanics
• Rehabilitation Engineering
• Human-Computer Interaction

Background:

• Assessing upper limb prosthesis (ULP) use in real-world settings is crucial for rehabilitation.
• Objective and inexpensive methods are needed to quantify functional upper extremity (UE) use in amputees.

Purpose of the Study:

• To adapt and validate a novel sensor-based method for identifying functional and non-functional UE use in upper limb amputees (ULA).
• To compare the efficacy of fixed-size versus variable-size data chunk analysis for classifying UE activity in ULA.

Main Methods:

• Five ULA and 10 control participants performed activities wearing wrist-mounted sensors (linear acceleration, angular velocity).
• Video data provided ground truth for annotating sensor data.
• Two analysis methods were employed: fixed-size and variable-size data chunk analysis using a Random Forest classifier.

Main Results:

• The fixed-size data chunk method achieved a median intra-subject accuracy of 82.7% and inter-subject accuracy of 69.8% for ULA.
• Variable-size data chunk analysis did not enhance classifier accuracy compared to the fixed-size method.
• The developed method demonstrates potential for objective UE use quantification in amputees.

Conclusions:

• The sensor-based method, particularly with fixed-size data chunk analysis, offers a promising, inexpensive tool for objectively assessing functional upper limb prosthesis use.
• This approach can aid in evaluating the effectiveness of upper extremity rehabilitative treatments for amputees.