Wired for Breath: The Quest to Predict Oxygen Consumption with Emg, Ins, and a Dash of Heart (Rate Variability)

Abstract

In recent years, estimating oxygen consumption (V˙O2) has gained significant attention due to its critical role in understanding metabolic rates and assessing physical activity levels. Direct measurement methods, while accurate, are impractical for continuous monitoring due to their bulky and restrictive nature. This has led to a growing need for reliable estimation techniques that enhance feasibility without compromising accuracy and precision. This paper proposes an approach using the Xception network to estimate V˙O2 by incorporating motion features, heart rate variability (HRV), and/or electromyography (EMG) signals. The approach was tested for oxygen consumption while walking and running on a level outdoor track. The lowest average estimation error of 3.41 ± 0.93 ml/kg/min was achieved using averaged EMG, HR, and motion features, and would be sufficient for tracking oxygen consumption of athletes and running enthusiasts. The lowest error using only data from EMG sensors was 5.93 ± 1.09 ml/kg/min, making this approach viable for continuous monitoring. Relying only on HRV data resulted in the lowest estimation error of 10.27 ± 2.93 ml/kg/min, which is too high for reliable V˙O2 monitoring.

Note:
Funding Information: This work was supported in part by the Academy of Finland, grants 287295 (under consortium “OpenKin: Sensor fusion for kinesiology research”) and 323472 (under consortium “GaitMaven: Machine learning for gait analysis and performance prediction”).

Declaration of Interests: The authors declare no conflict of interest.

Ethics Approval Statement: The Ethics Committee of the University of Jyväskylä approved the study. Participants were recruited between 5 May 2018 and 31 July 2018. All participants were informed about the content and purpose of the testing procedure, and provided written informed consent, witnessed by one researcher. The research was conducted in accordance with the World Medical Association Declaration of Helsinki.

Keywords: V˙O2estimation, electromyography (EMG), heart rate variability (HRV), in-
ertial navigation units (INU), neural networks, long short-term memory (LSTM)