Abstract: BACKGROUND: The tri-axial accelerometer is one of the most commonly used objective measurement tools for assessing physical activity volume, and its placing location is a key factor affecting its validity. Therefore, it is important to explore the impact of placing location on physical activity volume assessment for practical application.
OBJECTIVE: To investigate the variation of energy expenditure measured by tri-axial accelerometer ActiGraph wGT3X-BT in different positions of the body (wrist, waist, thigh, and ankle) and under different exercise conditions (4, 6, 8 km/h), thereby providing practical and theoretical reference for proper wearing and scientific use of wGT3X.
METHODS: Fifteen healthy male college students [age (24.8±1.7) years old, height (174.1±4.8) cm, body mass (69.5±8.6) kg, body fat percentage (18.1±5.8)%] volunteered to participate in the study. During the experiment, the subjects wore tri-axial accelerometers wGT3X (wrist, waist, thigh, and squat) and METALYZER-II lung function mask at the same time, and completed different exercise intensity on the sports treadmill (4 km/h normal walk, 6 km/h fast walk, 8 km/h medium speed running), to obtain acceleration count, estimated energy expenditure, and measured sports energy expenditure. The data were statistically analyzed using SPSS 19.0 version analysis software. 
RESULTS AND CONCLUSION: There was a significant difference in the estimated energy expenditure and the acceleration count measured by the accelerometer in the same placing position under three exercise intensities. When the running speed was 4 km/h, the tri-axial accelerometer at the wrist and waist slightly underestimated the energy expenditure of the exercise. At the speed of 6 and 8 km/h, the accelerometer at the wrist obviously underestimated the energy expenditure, while the accelerometer at the lumbar slightly overestimated the energy expenditure. The accelerometer at the thigh only showed a small error at 4 km/h, and then there was a marked overestimation trend. The accelerometer at the ankle joint dramatically overestimated the energy expenditure of the exercise under the three exercise intensities, and the measured energy expenditure and the acceleration count were the highest. Under the same exercise level, only the accelerometer at the lumbar had the estimated energy expenditure significantly correlated with the measured energy expenditure (moderately/highly correlated), which was the highest among the four placing positions. Moreover, the correlation between the acceleration counts of Y axis (sagittal axis) and Z axis (vertical axis) and vector magnitude and the measured energy expenditure was higher than that of the X axis (coronal axis). The correlation was in descending order: vector magnitude value > Z axis > Y axis > X axis. At the same time, the accelerometer at the waist showed the highest interpretation ability for the measured exercise energy expenditure, which was relatively stable with the smallest average error. To conclude, the accuracy of the estimated energy expenditure of the wGT3X is influenced by placing position and exercise intensity, the level of error increases with the increasing of exercise intensity. At low speeds (4 km/h), the error levels at the wrist, ankle and thigh are the smallest. Given wearing comfort and subject compliance, the three placing positions can be considered as alternatives to each other. The wGT3X accelerometer is best placed at the waist during walking and running as the accelerometer has the lowest error of the estimated energy expenditure and is the most stable in all tests.

Key words: tri-axial accelerometer, placing position, wGT3X, physical activity volume, treadmill