Physical activity classification using the GENEA wrist-worn accelerometer.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Physical Activity Recognition

Background:

• Accelerometer-based activity monitors are typically waist-worn for assessing habitual physical activity.
• Current monitors provide arbitrary counts, requiring classification into intensity levels.
• The Gravity Estimator of Normal Everyday Activity (GENEA) is an accelerometer-based device.

Purpose of the Study:

• To develop methods for classifying physical activities (walking, running, household, sedentary) using raw acceleration data.
• To compare the classification accuracy of a wrist-worn GENEA with a waist-worn GENEA.

Main Methods:

• Sixty participants performed 10-12 semistructured activities.
• Three GENEA accelerometers were worn: one at the waist, one on each wrist.
• Machine learning algorithms analyzed acceleration data (80 Hz) using FFT and wavelet decomposition for activity classification.

Main Results:

• The developed algorithm achieved high classification accuracy for both waist-worn (0.99) and wrist-worn GENEA (right wrist: 0.97, left wrist: 0.96).
• The algorithms successfully classified sedentary, household, walking, and running activities.

Conclusions:

• Algorithms were successfully developed for wrist-worn accelerometers to detect specific physical activities.
• Wrist-worn GENEA performance is comparable to waist-worn devices for physical activity assessment.