The Effects of ActiGraph Bandpass Filtering on Activity Counts During Continuous and Intermittent Lifestyle Activity

PURPOSE: The purpose of this study was to explore how increasing the upper limit of the bandpass filter frequency range affected accelerometer counts collected during treadmill walking and running, car driving and intermittent lifestyle activities METHODS: Part A included treadmill walking, running, and car driving (N=20) (mean ± [plus or minus] SD; age, 24.4±3.4 years; body mass index (BMI, 26.4±3.3 kg/m2 [kilograms per meter squared]). Part B included ten lifestyle activities ranging from sedentary behaviors to vigorous intensities (N=30) (mean±SD; age, 23.0±2.3 years; BMI, 25.1±3.8 kg/m2). Participants wore an ActiGraph accelerometer (GT3X+ in Part A and GT9X in Part B. on the hip. Participants completing Part B wore a Cosmed K4b2 [K4b squared]as a criterion measure of energy expenditure. Acceleration data were processed using a beta version of Actilife containing additional bandpass filter frequencies with upper limits of 5.0 Hz [Hertz] and 9.0 Hz, as well as, the 2.5 Hz default filter. Data were converted to 5-s epochs and the low frequency extension feature was employed. Cosmed data (VO2 [volume of oxygen] ml/min) were averaged over 30-s and then converted to relative VO2 (ml/kg/min) and metabolic equivalents (METs) for each activity. RESULTS: Part A: compared to the default bandpass filter, using a bandpass filter range of 0.25-9.0 Hz reduces the plateau effect seen during treadmill walking and running and significantly increases count values during car driving for all axes and vector magnitude. Part B: Increasing the bandpass filter frequency, significantly increased the count values on all axes during the lifestyle activities. Across all activities, the default filter had the strongest association between counts and METs, while the 5.0 Hz filter had the strongest association for lifestyle activities and the 9.0 Hz filter had the strongest association for locomotive activities. CONCLUSION: The plateau effect seen with the ActiGraph accelerometer can be reduced by increasing the bandpass filter frequency range. However, increasing the bandpass filter frequency range significantly increased the counts during car driving and lifestyle activities. Future work is needed to understand the impact that the increased count values will have on estimating energy expenditure