Circadia
  • Home
  • People
  • Publications
  • Projects
  • Tutorials
  • Join us/Visit
  • Design
  • Publications
  • Posts

ACTT Validation Study

From Movement to METs: Validating the ActTrust® actigraphy device

Research Pipeline
Actigraphy
R
Reproducible Research
A fully reproducible R/Quarto pipeline validating the ActTrust® device for energy expenditure and physical activity classification, benchmarked against the ActiGraph® GT3X+ and indirect calorimetry.
Published

September 1, 2024

About

This repository contains all data and analysis code for a controlled laboratory validation of the ActTrust® actigraphy device (Condor Instruments, São Paulo, Brazil). The study compares ActTrust® activity counts to those from the ActiGraph® GT3X+ accelerometer, and relates both to metabolic equivalents (METs) measured by indirect calorimetry during a treadmill protocol in healthy young adults.

The entire analysis is implemented as a single Quarto notebook (index.qmd) that renders a self-contained HTML report with all figures, tables, and statistical outputs inline — making every step fully auditable and reproducible.

Batista ES, Basilio Gomes SR, Morais Ferreira AB, França LGSF, Fontenele Araújo J, Mortatti AL, Leocadio-Miguel MA. From Movement to METs: A Validation of ActTrust® for Energy Expenditure Estimation and Physical Activity Classification in Young Adults. bioRxiv 2025. doi:10.1101/2025.05.16.654458

Study Design

56 healthy young adults (34 men, 22 women; aged 18–35) completed a controlled treadmill protocol in a single laboratory session. Participants simultaneously wore four devices — two ActiGraph® GT3X+ and two ActTrust® units at hip and wrist — while oxygen uptake was measured breath-by-breath by indirect calorimetry (Quark CPET, Cosmed®).

Condition Speed Description
Rest — Seated, no movement
Walking 3 km/h Slow walk
Walking 5 km/h Brisk walk
Walking/Running 7 km/h Fast walk / slow run
Running 9 km/h Running

Each condition lasted 10 minutes, separated by 5-minute rest periods. Activity counts and METs were averaged over the central 4 minutes of each condition to exclude transition effects.

Key Contributions

  • First published cut-points for physical activity intensity classification using ActTrust® devices at hip and wrist placement
  • General linear model for estimating energy expenditure (METs) from ActTrust® counts, with device × placement interaction terms
  • Balanced accuracies above 0.77 for all intensity ranges, exceeding 0.90 for light and moderate activity
  • Bland-Altman analysis of predicted vs measured METs per device × placement
  • Sex-stratified supplementary analysis addressing body composition differences

Derived Cut-points

Physical activity intensity cut-points derived by inverting the regression model at MET thresholds of 3, 6, and 9 (counts/min):

Device (Placement) Light < 3 METs Moderate 3–6 METs Vigorous 6–9 METs
ActTrust® hip < 5057 5057–23339 23339–46410
ActTrust® wrist < 3761 3761–22368 22368–47203
GT3X+ hip < 1132 1132–4853 4853–9468
GT3X+ wrist < 1698 1698–9503 9503–19787

Reproducibility

The analysis can be reproduced in two ways:

RStudio — open ACTT_validation_study.Rproj and render index.qmd.

Docker — docker-compose up builds the full R environment and renders the notebook automatically.

Data are archived on Zenodo: doi:10.5281/zenodo.15315735

Links

  • 💻 GitHub
  • 📄 Preprint (bioRxiv)
  • 🗄️ Data (Zenodo)
 

Made with ❤️ and Quarto. © 2026. This work is openly licensed via CC BY 4.0