Aliasing affects ActiLife software raw accelerometry to count conversion from different sampling frequencies

Area of Science:

• Biomedical Engineering
• Physical Activity Measurement
• Signal Processing

Background:

• Accelerometers are crucial for objective physical activity quantification.
• ActiGraph accelerometers record acceleration signals at various sampling frequencies (fs).
• ActiLife software may compute additional counts from signals with fs>30 Hz compared to default fs=30 Hz.

Purpose of the Study:

• Investigate the origin of erroneous accelerometry counts in ActiLife software.
• Identify the cause of count overestimation with higher sampling frequencies.
• Recommend an adjusted method to ensure accurate physical activity data.

Main Methods:

• Generated synthetic piecewise-frequency sinusoidal signals (0-15 Hz) at fs=30, 50, and 100 Hz.
• Resampled artificial acceleration raw signals to 30 Hz using different antialiasing lowpass filters.
• Computed ActiLife counts and analyzed the impact of aliasing on the results.

Main Results:

• Improperly attenuated aliasing replicas induced by antialiasing filters caused spurious frequencies within the ActiLife bandpass filter.
• Fictitious accelerometry counts were reproduced for signals around 10 Hz.
• Count overestimations at fs=50 and 100 Hz were attributed to aliasing within the ActiLife count filter's frequency bandwidth.

Conclusions:

• Aliasing due to inadequate antialiasing filtering is the primary cause of accelerometry count overestimation in ActiLife software at higher sampling frequencies.
• Adjusting antialiasing filter parameters prior to ActiLife processing can prevent erroneous counts.
• Accurate physical activity data can be ensured through appropriate signal preprocessing or advanced mathematical programming.