From motion to metabolism: investigating the relationship between accelerometer and VO2metrics across five age groups for optimal calibration of physical activity intensity.

Abstract

Objective.The aim was to examine the relationship between accelerometer and oxygen consumption (VO₂) metrics and to what extent the metrics are normalized across age and body size, to allow a single calibration regression line for absolute physical activity (PA) intensity.Approach.Hip-mounted accelerometer data and VO₂measurements were collected from 51 participants across five age cohorts (4-5; 6-8; 10; 15 and 20 years) during resting, walking and running on a treadmill in laboratory setting. Linear regressions were used to determine four accelerometer metrics' (AG, 4 Hz frequency extended method (FEM), 10 Hz FEM and Euclidean norm minus one) contribution to explained variance (adjustedR²) in six VO₂metrics (VO₂, VO₂/kg¹, VO₂/kg^0.67, VO₂/kg^0.75, MET_measuredand MET_fixed). Plots were generated for visual representations together with log-linear regression, finding the optimal scaling exponent for VO₂.Main result.10 Hz FEM explained the highest amount of explained variance when related to VO₂/kg^0.75, 92.4%, with minimal remaining between-group and inter-individual variance. The relationship demonstrated a linear shape. The most used accelerometer metric, AG counts, together with traditionally used reference standard, MET_fixed, show substantially lower explained variance, 60.2%, with large between-group and inter-individual variance, insufficiently adjusting for physiological and biomechanical variability. The best body weight scaling factor for VO₂was 0.77. Findings support the use of a single linear calibration regression line for absolute PA intensity across wide-ranging age-groups, accounting for biomechanical and physiological variance.Significance.This enables reliable and meaningful comparisons of PA intensity across age-groups, possibly also across childhood into adulthood, overcoming traditional limitations and enhancing research quality.