Fig. 3
Linear power calibration. Using our power calibration dataset, we considered two linear models to calibrate vectorial dynamic body acceleration (\({a}_{d}\), in m/s2) using allometric estimates of energy expenditure (\(\dot{E}\), in W) during resting (Eq. 1) and locomotion (Eq. 3). In the ACTIWAKE model (in pink; smaller slope and larger intercept), we performed linear regression using data from locomotion (walking, running) while for the ACTIREST24 model (in blue; larger slope and smaller intercept), we used data from both locomotion and resting