Activity classification using the GENEA: optimum sampling frequency and number of axes

Area of Science:

• Wearable technology
• Biomedical engineering
• Signal processing

Background:

• The GENEA wearable sensor accurately classifies activities like walking and running at 80 Hz on three axes.
• Optimal sampling frequency and number of axes for accurate activity classification are not fully understood.

Purpose of the Study:

• To evaluate the impact of reduced sampling rates (5-80 Hz) and axes (1-3) on GENEA activity classification accuracy.
• To determine if lower data collection parameters affect the classification of sedentary, household, walking, and running activities.

Main Methods:

• Sixty participants performed semi-structured activities wearing a GENEA accelerometer.
• Data from single, dual, and three axes were analyzed at sampling rates of 5, 10, 20, 40, and 80 Hz.
• Mathematical models extracted features for activity classification.

Main Results:

• High classification accuracy (94.98%–97.4%) was maintained across sampling rates of 10-80 Hz, regardless of the number of axes.
• Accuracy slightly decreased only at the lowest sampling rate of 5 Hz.

Conclusions:

• Sampling frequencies above 10 Hz and using more than one axis do not improve activity classification accuracy.
• Lower sampling rates and single-axis measurements reduce data load and improve processing efficiency.
• Further research is needed to confirm these findings with a broader range of activities.