What are the main findings? a) Optimization of the simulated oxygen uptake alone still resulted in both large overestimations and underestimations of total energy expenditure and aerobic-anaerobic energy supplies. b) Only when combined with the corrected calculation of the aerobic energy supply, no statistically significant differences were found for total energy expenditure and aerobic supply, except during repeated sprints. What are the implication of the main findings? a) The anaerobic energy supply is still largely overestimated, thus the calculation of the metabolic power itself should be reconsidered. b) Even more precise results concerning the distinction between aerobic-anaerobic energy supplies could be achieved by including physiological data, namely heart rate, into the simulation of oxygen uptake. Abstract: In intermittent sports, tracking technologies are commonly used to monitor external and internal loads. The metabolic power model solely uses speed and acceleration data to simulate metabolic power, oxygen uptake, energy expenditure, and aerobic-anaerobic supplies. This study aimed to improve the simulation of oxygen uptake within the metabolic power model, thereby increasing its validity to estimate metabolic loads during intermittent running. Twelve male athletes (24 ± 3 years) performed different intermittent running-based exercises. These data were previously collected and used for secondary analysis within this study. The simulation of oxygen uptake was optimized by different approaches: (i) formerly detected bias (Offset model), (ii) data-driven modeling using differential evolution (Mongin model), and (iii) correction of the aerobic supply calculation. The simulations were compared to the measured oxygen uptake via a portable respiratory gas analyzer and the resulting metabolic loads to those derived by the established 3-component model. For statistical analysis, one-way repeated measures ANOVA or Friedman test with corresponding effect sizes were applied. Overall, the Mongin model demonstrated the best predictive accuracy (MAE = 4.99 ± 1.12 mL/min/kg) compared to measured oxygen uptake and, combined with the corrected calculation, total energy expenditure and aerobic supply did not significantly differ to the standard (p = 0.056; trivial to large effect sizes). In conclusion, our optimizations reduce discrepancies of the tracking-based metabolic power model regarding total energy expenditure and aerobic supply compared to the established 3-component model.
© Copyright 2025 Sensors. Alle Rechte vorbehalten.