Classifying household and locomotive activities using a triaxial accelerometer.

Area of Science:

• Biomechanics
• Wearable Technology
• Data Science

Background:

• Distinguishing between household and locomotive physical activities is crucial for health monitoring and activity recognition.
• Existing methods may lack accuracy or require complex sensor setups.

Purpose of the Study:

• To develop and validate a novel algorithm for classifying physical activity into household or locomotive categories.
• To utilize data from a triaxial accelerometer for accurate activity classification.

Main Methods:

• Sixty-six volunteers performed 12 distinct physical activities in a laboratory setting.
• A triaxial accelerometer recorded movement data, which was filtered using a second-order Butterworth high-pass filter (0.7 Hz cutoff).
• Ratios of unfiltered to filtered total acceleration (TAU/TAF) and filtered vertical to horizontal acceleration (VAF/HAF) were calculated for classification.

Main Results:

• The developed algorithm using the TAU/TAF cut-off value achieved an average correct discrimination rate of 98.7% in the cross-validation group.
• An alternative method using VAF/HAF showed high accuracy but had a lowest correct discrimination of 63.6% for specific activities.

Conclusions:

• The novel algorithm employing the TAU/TAF cut-off value demonstrates high accuracy in classifying household and locomotive physical activities.
• This algorithm provides a robust and efficient tool for objective physical activity assessment using accelerometer data.