武汉体育学院

The aim was to examine whether pre-session well-being indicators differentially predict physiological (% heart-rate [HR] reserve; %HRreserve) versus perceived (rating of perceived exertion; RPE) internal load across soccer training formats. Seventy-seven male soccer players (20.6 +/- 1.6 y) completed four monitored sessions over two weeks: 1v1 small-sided games, 5v5 small-sided games, long high-intensity interval training (HIIT), and short HIIT. Mean %HRreserve was obtained from heart-rate monitoring, and RPE was collected immediately post-session. Linear mixed-effects models estimated associations of well-being, training format, and well-being and format interactions with each internal-load outcome. No well-being indicator showed a clear association with %HRreserve. For RPE, longer sleep duration was associated with lower perceived exertion (B = -0.42, p = 0.007), whereas greater musculoskeletal pain was associated with higher perceived exertion (B = 0.47, p = 0.012). A significant sleep and short-HIIT interaction (B = 0.79, p = 0.012) indicated that the sleep-RPE association differed between short HIIT and the reference format (1v1). In this sample, sleep duration and musculoskeletal pain were associated with perceived, but not HR. These findings support interpreting RPE, HR-derived measures, and brief well-being monitoring as complementary tools in soccer training contexts, while causal inference and broader generalisation remain limited.

Sports broadcasters invest heavily in interactive features on social media, assuming that functional engagement (likes, shares, comments) builds viewer loyalty. This study challenges that assumption. Using Partial Least Squares Structural Equation Modeling (PLS-SEM) on survey data from 450 active sports viewers, we test a model that separates the effects of functional, emotional, and communal engagement on loyalty, mediated by social presence and channel commitment. Results reveal that functional engagement significantly enhances social presence but fails to translate into loyalty, as higher perceived social presence is associated with lower channel commitment. Instead, viewer loyalty is strongly and significantly driven by channel commitment, which itself is primarily fostered by emotional connection and, to a lesser extent, communal engagement. Theoretically, this study demonstrates the primacy of affective bonds over transactional interactions in the contemporary sports media economy and provides empirical evidence that questions the utility of Social Presence Theory as a direct predictor of loyalty in this context. The findings suggest a critical re-evaluation of how audience engagement is conceptualized and measured by both scholars and practitioners.

Brief pre-class physical exercise may enhance executive function (EF) processes proximal to mathematics learning. However, the effect is likely dependent on the alignment between the physical exercise and subsequent instructional demands. We tested a dual-process framework in authentic classrooms, predicting selective gains in mathematics-specific inhibitory control rather than broad executive function. Using a cluster-randomized trial, four Grade-5 classes (N = 182 students) were assigned to mentally passive sedentary behavior (e.g., video watching), mentally active sedentary behavior (e.g., numeracy practice), 5-minute physical exercise, or 10-minute physical exercise. EF outcomes were assessed at baseline, immediately post-intervention, and after the mathematics instruction using a domain-general flanker task and a mathematics-specific negative priming task. Both physical exercise groups outperformed mentally passive sedentary behavior on the inverse efficiency score of the negative priming task at post-instruction, with no differences between 5- and 10-minute physical exercise, and no consistent advantages on the flanker interference performance. These results support a task–context alignment mechanism in which brief physical exercise may prime motivational/alerting states that translate into improved inhibition when followed by domain-relevant instruction. Short, pre-class physical exercise routines can be integrated without displacing teaching time and may enhance readiness for mathematics learning by selectively improving mathematics-specific inhibitory control. We discuss implementation parameters for classroom use and implications for theory on how physical exercise influences psychological processes central to learning.

Foundation models are transforming transcriptome analysis, yet most RNA sequencing (RNA-seq) foundation models are pretrained on sparse single-cell RNA-seq data, which typically captures only ∼3,000 genes per cell. This creates a need for models tailored to bulk transcriptomes, a modality that profiles ∼16,000 genes per sample and supports clinical and tissue-level analyses. Here, we present BulkFormer, a foundation model for bulk transcriptome analysis. BulkFormer contains ∼150 million parameters, covers 20,010 protein-coding genes, and is pretrained on 581,503 human bulk RNA-seq profiles. Its hybrid encoder combines a graph neural network to model explicit gene-gene relationships with a Performer module to capture global expression dependencies. Across five downstream tasks, BulkFormer outperformed existing single-cell foundation models while requiring substantially lower training cost. These results identify pretraining data modality as a key determinant of foundation model performance and establish BulkFormer as a framework for bulk transcriptome modeling. A record of this paper's transparent peer review process is included in the supplemental information.