Abstract

The purposes of this study were to recalibrate a new motion sensor wristband (ViM sports memory: ViM) to metabolic equivalents (METs) during walking and running, and to examine the cross-validity of the recalibration model. Forty-five healthy Japanese (23 males and 22 females) voluntarily participated in this study. The participants were divided into two groups: a recalibration group (16 males and14 females) and a cross-validity group (7 males and 8 females). The participants performed 5 min of treadmill walking at 3.6, 4.8, and 6.0 km/h, and treadmill running at 7.2, and 9.6 km/h. The ViM was placed on the non-dominant wrist. Simultaneous measurements of the ViM and an indirect calorimeter (IC) were continuously recorded during all exercises. The recalibration models were calculated for males and females from data of the recalibration group using general linear mixed model. The cross-validity was examined by correlation coefficients, repeated ANOVA, and paired t-test adjusted by Bonferroni's method. In the recalibration group, the recalibration models showed high determination coefficients (male: R2=0.74, SEE=1.51 METs; females: R2=0.84, SEE=1.16 METs). In the cross-validation group, there were significant correlation coefficients at 3.6, 4.8, 6.0, and 9.6 km/h (r>=0.56, p<0.001). Differences (−4 to 13%) between the IC measurements and the recalibration model estimations were not significant (p>0.185) at any speed. These results showed that the recalibration model improved the validity and accuracy of the ViM. Verf.-Referat