Continuous Monitoring of Muscle Oxygenation in Endurance Athletes During Incremental Cycling: Experimental Validation of a Wearable Continuous-Wave NIRS Sensor Using Frequency-Domain Near-Infrared Spectroscopy

TL;DR

This study validates a wearable sensor for measuring muscle oxygen levels during cycling, showing it performs reliably compared to a lab device.

Contribution

The study experimentally validates a wearable CW-NIRS sensor against FDNIRS for muscle oxygenation monitoring in real-world exercise settings.

Findings

Strong correlations (r = 0.792) were found between the wearable and lab devices for muscle oxygen saturation measurements.

Most subjects showed an RMSD < 5% and a mean bias of 0.005 between the two devices.

Bland–Altman analysis showed 95% of measurements fell within −8.1% and 7.6% SmO2.

Abstract

Individuals often lack field-based tools to monitor exercise effectiveness. New sensing methods may allow for an improved measurement of the individualized response to exercise by monitoring oxygen kinetics directly in muscle tissue. This study aimed to validate a non-invasive wearable sensor capable of measuring muscle oxygen saturation (SmO2) using continuous-wave near-infrared spectroscopy (CW-NIRS) against a laboratory-validated frequency-domain NIRS (FDNIRS) device. Ten physically fit adults performed an incremental cycling test until voluntary exhaustion. Devices were placed on contralateral rectus femoris muscles. SmO2 was simultaneously measured continuously for the duration of the protocol. Time series alignment was performed using linear interpolation to enable direct comparison between devices at matched time points. Z-score normalization accounted for inter-individual differences in a group-level analysis. Individual subject validation showed strong correlations between the two devices (r = 0.792, range: 0.69–0.88, p < 0.001) with an RMSD < 5% for most subjects, a mean bias of 0.005 and low proportional bias (−0.199) between all paired measurements. Group-level analysis demonstrated a correlation of r = 0.788. Bland–Altman analysis revealed that 95% of all measurements fell between −8.1% and 7.6% SmO2. The CW-NIRS device delivered reliable performance compared to the FDNIRS device, offering potential applications for real-time physiological monitoring during exercise and performance assessment.

Figures