Related Experiment Video
Updated: Feb 28, 2026
11:21
Methodology for Establishing a Community-Wide Life Laboratory for Capturing Unobtrusive and Continuous Remote Activity and Health Data
Published on: July 27, 2018
8.9K
Extracting aerobic system dynamics during unsupervised activities of daily living using wearable sensor machine
Thomas Beltrame1,2, Robert Amelard3,4, Alexander Wong3,4
1Faculty of Applied Health Sciences, University of Waterloo , Waterloo, Ontario , Canada.
Journal of Applied Physiology (Bethesda, Md. : 1985)
|June 10, 2017
Summary
Wearable sensors can now assess aerobic system dynamics during daily activities using predicted oxygen uptake (V̇o2). This technology offers potential for early detection of health changes in real-world settings.
Area of Science:
- Exercise Physiology
- Biomedical Engineering
- Machine Learning in Healthcare
Background:
- Current physical activity algorithms lack insight into the physiological systems supporting energy supply.
- Longitudinal analysis of oxygen uptake (V̇o2) using wearables could enable practical tools for studying aerobic system dynamics.
- Assessing V̇o2 kinetics in real-world settings is crucial for understanding functional health.
Purpose of the Study:
- To evaluate aerobic system dynamics using predicted V̇o2 data from wearable sensors during unsupervised activities of daily living (μADL).
- To explore the feasibility of using machine learning and wearable data for V̇o2 prediction and analysis.
- To investigate the correlation between predicted V̇o2 dynamics during μADL and laboratory-measured V̇o2 kinetics.
Main Methods:
- Thirteen healthy men wore hip accelerometers, heart rate monitors, and respiratory bands during ≈6 hours/day for 4 days of μADL.
- A random forest regression model processed sensor data to predict V̇o2.
- Frequency-domain analysis of accelerometer and predicted V̇o2 data was performed on active samples (ACCHIP > 0.05 g).
Main Results:
- Temporal characteristics of predicted V̇o2 during μADL significantly correlated with measured V̇o2 during laboratory protocols (R²=0.82, P < 0.001).
- The study successfully demonstrated the use of wearable sensors and machine learning for estimating aerobic system dynamics outside controlled environments.
- Optimal frequency domain analysis was achieved by selecting active periods during μADL.
Conclusions:
- Aerobic system dynamics can be effectively investigated during unsupervised daily activities using wearable sensors and advanced algorithms.
- These predictive algorithms hold potential for integration into wearable systems for early detection of health status changes.
- This approach may contribute to models of functional health and guide personalized healthcare by identifying subtle impairments in aerobic response.
