Supporting data for "Lower-limb mechanical power accounts for running energy expenditure and enables single-IMU estimation"

This repository contains data and code associated with the following manuscript: [Title] Lower-limb mechanical power accounts for running energy expenditure and enables single-IMU estimation [Authors] Jinsung Jung, Hyerim Lim, Sukyung Park This study investigated segmental and planar contributions to whole-body joint mechanical power during running, and developed an IMU-based energy expenditure (EE) estimation framework using biomechanically informed intermediate variables. Eight healthy adult male recreational runners performed treadmill running at five speeds (2.5, 2.8, 3.1, 3.4, 3.7 m/s). Whole-body kinematics (motion capture), ground reaction forces (instrumented treadmill), sacrum IMU signals, and metabolic data (indirect calorimetry) were collected simultaneously. Joint dynamics were computed in Visual3D v6 and used to derive efficiency-weighted mechanical power per body segment. A single sacrum-mounted IMU was used to estimate stance-phase joint dynamics via an artificial neural network (ANN), which were then converted to estimated EE using segment-specificscale factors. Walking data from a prior dataset (Jung et al., Sci. Rep., 2025) were included for comparison of segmental and planar joint power patterns between gaits.