Epoch length and accelerometer outputs in children: comparison to direct observation

Area of Science:

• Pediatric physical activity monitoring
• Wearable sensor technology
• Exercise physiology in children

Background:

• Accurate measurement of physical activity in children is crucial for public health.
• ActiGraph accelerometers are widely used but require careful calibration.
• Understanding the impact of device settings on data accuracy is essential for research.

Purpose of the Study:

• To evaluate how epoch length and activity count cutpoints affect ActiGraph accelerometer estimates of moderate-to-vigorous physical activity (MVPA).
• To compare ActiGraph MVPA estimates with direct observation (DO) in fifth-grade children during physical education (PE).

Main Methods:

• Thirty-two fifth-graders wore ActiGraph accelerometers during a 30-minute PE class.
• Direct observation using C-SOFIT served as the criterion standard for MVPA.
• ActiGraph data were analyzed using various epoch lengths (5-60s) and three cutpoint methods (AG_T, AG_M, AG_F).

Main Results:

• The Freedson et al. (AG_F) cutpoint provided estimates comparable to direct observation (93-100% detection).
• Treuth et al. (AG_T) and Mattocks et al. (AG_M) cutpoints significantly underestimated MVPA (26-61% detection).
• Increasing epoch length with AG_F showed modest increases in RMSE, with 5-second epochs yielding the lowest error.

Conclusions:

• ActiGraph accelerometers using the AG_F cutpoint accurately estimate mean MVPA in children during PE across different epoch lengths.
• Shorter epoch lengths (e.g., 5 seconds) are recommended to minimize error in individual MVPA estimates.
• Both epoch length and activity count cutpoint selection are critical for precise MVPA detection in children.