Calibration of an accelerometer to measure physical activity in preschool children: A feasibility study

Physical activity measurements are essential for the evaluation and prevention of childhood obesity. Although physical activity (PA) data are currently collected with accelerometry, using the existing accelerometers had limitation with cost efficiency issues. The Best Fit Friend (BFF), a new accelerometer PA monitor system, was develop to solve the cost issue; however, validity of BFF to measure the PA in preschool children is still not clear. The primary purpose of this thesis was to examine the feasibility of BFF accelerometer for measure 3 to 5 year-olds preschoolers’ PA measurement. Based on the parental response and participant’s feedback, research and device feasibility were estimated. The second purpose was to validate the BFF accelerometer though a comparison with ActiGraph. Pearson correlation showed a moderate-high and/or high correlation between BFF accelerometer and ActiGraph. The third purpose was to clarify the difference of attachment placement: wrist and waist. Wilcoxon signed ranks test was to examine the statistical difference between dominant hand and non-dominant hand (Z = 4.99, p < .000, d = -.481), whereas there was not statistical differ with left wrist and waist, back and waist (Z = -1.39, p < .165, d = .130; Z = -.529, p < .597, d = -.050). The fourth purpose was to calibrate the BFF accelerometer to establish cut-off points. ROC curve analysis was performed. The last purpose was to decide the best placement of BFF accelerometer for children’s PA measurement. Based on comparison between AUC and cut-off, waist and back placements was the best BFF location, and right wrist, waist, and back conduct the best cut-off points. Overall, the BFF is a valid device to measure PA in preschool children ranged from 3 to 5 years old