Tapping The Potential Presented By The Gravity Component Of An Accelerometer Signal

PURPOSE: In order to quantify the acceleration due to physical activity from an accelerometer signal, the gravity component of acceleration is typically removed from the signal. Yet, when inactive, the orientation of the gravity vector of a triaxial accelerometer provides valuable information, particularly when the monitor is worn on the wrist as utilised in the Sedentary Sphere. We aimed to further explore use of the gravity vector to 1) identify accelerometer wear-site and 2) classify posture using the Sedentary Sphere in data from two brands of wrist-worn accelerometer. METHODS: 1) Wear-site: 22 children, aged 10-12 y, wore a GENEActiv at the wrist and at the hip for 7 days. The angle between the forearm and vertical for the wrist-worn monitor and between the pelvis and the vertical for the hip-worn monitor was calculated for each 5 s epoch. The standard deviation of this angle (SDangle) was calculated over time for windows of varying lengths (1-720 min). We hypothesised that the wrist angle would be more variable than the hip angle. 2) Posture: 34 adults, aged 20-40 y, wore a GENEActiv and an ActiGraph GT3X+ on their non-dominant wrist and an activPAL3 on their thigh for 24 h. Posture (sitting/lying vs upright) was estimated based on the orientation of the gravity vector for both the GENEActiv and ActiGraph and compared with the activPAL. RESULTS: 1) Wear-site: Wear site could be discriminated based on SDangle; the shorter the time window the lower the optimal threshold and area under the receiver-operating-characteristic curve (AUROC) for discrimination of wear-site (AUROC = 0.833 (1 min) - 0.952 (12 h)). Classification accuracy was good for time windows >15 min (sensitivity > 92%, specificity > 88%, AUROC > 0.93). 2) Posture: Time estimated sitting/lying was 534±144 min/d (activPAL), 523±143 min/d (GENEActiv) and 528±137 min/d (ActiGraph). Strong intra-class correlations were evident between the activPAL and accelerometer posture classifications, irrespective of brand (>0.93, 95% confidence interval ≈ 0.84-0.97). CONCLUSION: The gravity component of the acceleration signal of a triaxial accelerometer provides valuable information on the orientation of the monitor. This can be exploited to distinguish between hip and wrist wear-sites and to estimate posture from a wrist-worn accelerometer, independent of brand.