Yingzhe Song

This study examined whether acute exercise performed within individualized physiological intensity zones affects multidimensional soccer shooting performance. Twenty male collegiate soccer players completed a Yo-Yo Intermittent Recovery Test Level 1 with portable gas analysis to determine the ventilatory threshold (VT) and respiratory compensation point (RCP). Three individualized zones were defined: Low (RCP). In a randomized design, players completed three 3 min shuttle-running bouts, each followed immediately by the 356 Soccer Shooting Test. Ball velocity (BV), shooting accuracy (SA), and shooting quality (SQ) were analyzed using repeated-measures ANOVA. Exercise condition significantly affected SA (p = 0.013) and SQ (p = 0.007), but not BV (p = 0.216). Bonferroni-adjusted comparisons showed that SA and SQ were lower in High than in Low, whereas no pairwise BV comparison reached significance. A sensitivity analysis using all ten recorded attempts rather than the original best-seven scoring approach showed an overall condition effect for BV without a significant pairwise comparison, retained overall effects for SA and SQ, and showed that the Low–High contrast remained robust only for SQ. Baseline comparisons were not significant. These findings indicate condition-specific shooting responses, with the clearest evidence for lower SQ after High compared with Low, supportive evidence for lower SA, and no significant primary condition effect for BV.

Dynamic oxygen uptake (VO₂) reflects moment-to-moment changes in oxygen consumption during exercise and underpins training design, performance enhancement, and clinical decision-making. We tackled two key obstacles—the limited fusion of heterogeneous sensor data and inadequate modeling of long-range temporal patterns—by integrating wearable accelerometer and heart-rate streams with a convolutional neural network–LSTM (CNN-LSTM) architecture and optional attention modules. Physiological signals and VO₂ were recorded from 21 adults through resting assessment and cardiopulmonary exercise testing. The results showed that pairing accelerometer with heart-rate inputs improves prediction compared with considering the heart rate alone. The baseline CNN-LSTM reached R² = 0.946, outperforming a plain LSTM (R² = 0.926) thanks to stronger local spatio-temporal feature extraction. Introducing a spatial attention mechanism raised accuracy further (R² = 0.962), whereas temporal attention reduced it (R² = 0.930), indicating that attention success depends on how well the attended features align with exercise dynamics. Stacking both attentions (spatio-temporal) yielded R² = 0.960, slightly below the value for spatial attention alone, implying that added complexity does not guarantee better performance. Across all models, prediction errors grew during high-intensity bouts, highlighting a bottleneck in capturing non-linear physiological responses under heavy load. These findings inform architecture selection for wearable metabolic monitoring and clarify when attention mechanisms add value.