Author(s): DE GIOVANNI, E., TEIJEIRO, T., Institution: BASQUE CENTER FOR APPLIED MATHEMATICS (BCAM), Country: SPAIN, Abstract-ID: 2570

INTRODUCTION:
Monitoring oxygen consumption (VO2) during maximal exercise testing typically relies on gas exchange analysis, while indirect estimations rely on standard wearable sensors, with a primary focus on heart rate (HR) and RR intervals dynamics [1, 2]. However, recent evidence suggests a correlation between VO2 and other electrocardiogram (ECG) features, specifically those related to cardiac repolarization and T-wave morphology [3]. In this work, we investigate the trend of the T-wave amplitude relative to the R-peak as a predictor for the maximal oxygen uptake, hypothesizing that T-wave amplitude displays a structural breakpoint coinciding with the beginning of the VO2 plateau at exhaustion.
METHODS:
Data were collected from 20 subjects (13 men, 7 women; age range: 20–37 years) performing maximal graded exercise tests on a cycle ergometer [4]. Both gas exchange and single-lead ECG data (Biopac, 250 Hz) were gathered, with physiological experts identifying ventilatory thresholds (VT1, VT2) and VO2max. We focused on segments from VT2 to the test conclusion (including 3 min of recovery), assuming that two breakpoints would occur corresponding to the VO2 plateau starting point and the recovery onset (exhaustion). We annotated the R peaks using our automated algorithm (Bayeslope) [4] and revised them manually. Then, we delineated the T-wave using our established method [5] and extracted the amplitudes relative to the corresponding R peak. To capture the trend of the T amplitude and smooth the time series, we applied different rolling windows in the range 30-165 s. Finally, we applied a piecewise linear regression [6] to determine the breakpoints where the gradient changes.
RESULTS:
Our analysis shows that the absolute distance from the expert-identified VO2 plateau was under 65, 35, and 15 seconds for 80% (n=16), 50% (n=10), and 30% (n=6) of the cohort, respectively. Among the 16 subjects considered acceptable (<65 seconds), the 97.5% confidence interval for this distance was below 6 seconds.
CONCLUSION:
ECG features, particularly T-wave morphology, hold promise for automatically monitoring and estimating the VO2 plateau during maximal exercise. This method may also be extended to analyze VT1 and VT2, enhancing VO2 trend monitoring capabilities with ECG-based wearable sensors. The inter-subject variability observed necessitates further analysis of T-wave dynamics and other ECG features beyond HR to develop comprehensive fitness profiles and personalized monitoring.
[1] Neshitov, 2023;[2] Shandhi, 2020;[3] Veselik, 2024;[4] De Giovanni, 2022; [5] De Giovanni, 2020;[6] Pilgrim, 2021