University of Applied Sciences Kufstein

Table tennis has increasingly been adopted as a tool to promote physical and mental health, yet evidence on its outcomes and implementation remains scattered. This study conducted a rapid scoping review to summarise available research on the health outcomes of table tennis within recreational or non-elite settings and identify how table tennis-for-health activities are structured and delivered. Peer-reviewed articles in English were included when they focused the outcomes of table tennis participation on health in community or social settings. Searches across two multidisciplinary databases, complemented by reference screening, led to 17 studies published between 2010 and 2025 being included. Studies were then charted for their methodological, intervention and outcome characteristics. Most studies employed quantitative methods, with experimental or controlled designs predominating, and targeted children, adolescents, older adults, and individuals with conditions such as ADHD or Parkinson’s disease. Across various settings, table tennis was associated with improvements in physical fitness, balance, agility, and body composition, alongside cognitive benefits such as enhanced executive functioning and visual–perceptual skills. Psychological and social outcomes, including improved self-efficacy, emotional regulation, cooperation and social interaction, were also reported. Though no formal quality assessment was conducted, there are clear methodological limitations, including small sample sizes, geographic and gender imbalances, and limited reporting on intervention characteristics that restrict the strength and generalisability of the findings. Overall, this review provides a starting point for trainers and health professionals in the area, presenting promising but preliminary evidence for table tennis as a health-enhancing activity and highlighting the need for more rigorous and comprehensive evaluation.

Self-supervised data splitting has emerged as a promising paradigm for sparse-view CT reconstruction, enabling training from incomplete measurements without fully sampled ground truth. However, the influence of key design choices, including partitioning strategy, preprocessing, and inference, remains insufficiently understood. In this work, we introduce a unified framework that decomposes splitting-based reconstruction into these three components, enabling controlled comparison of existing methods and two incremental extensions: multi-partition splitting and an alternative inference strategy. Experiments on simulated LoDoPaB-CT data under independent and correlated noise, together with validation on the real-world 2DeteCT dataset, show that the optimal partitioning strategy strongly depends on the measurement noise structure. Lattice-based splitting performs favorably under independent noise, whereas angular masking is more robust under correlated noise and real measured data. Multi-partition splitting consistently improves over pure projection-wise splitting in several settings. Complementary perceptual and structural metrics, including LPIPS and HaarPSI, reveal differences between masking strategies that are less apparent from PSNR and SSIM alone. These results provide practical guidelines for designing self-supervised sparse-view CT reconstruction methods and highlight the limitations of common independence assumptions in realistic imaging environments.

This study explores the evolving discourse within the Sport for Development (SFD) field through an analysis of over 10,000 English-language articles published on the sportanddev platform between 2003 and 2024. Positioned as a central hub for SFD communication, this research examines how the agenda is constructed within the sportanddev platform, with a particular focus on the organizations, topics, goals, and target groups of the articles. The findings reveal a significant decline in article volume post-2016, coinciding with the closure of the UN Office on Sport for Development and Peace. Despite the platform’s open nature, content is disproportionately produced by a small number of well-resourced international organizations, suggesting persistent power imbalances. Football dominates as the most featured sport worldwide, but there are also strong regional preferences. Thematic focuses such as youth development, gender, and disability fluctuate over time, often aligning with global events like the Paralympics or the COVID-19 pandemic. This suggests that external events and well-resourced organisations still largely drive the agenda in SFD. Overall, this study highlights how digital platforms such as sportanddev both reflect and reinforce existing hierarchies within SFD, while also offering potential for more inclusive and critical engagement.

Solving image reconstruction problems of the form 𝐀𝐱 = 𝐲 remains challenging due to ill-posedness and the lack of large-scale supervised datasets. Deep Equilibrium (DEQ) models have been used successfully but typically require supervised pairs (𝐱,𝐲). In many practical settings, only measurements 𝐲 are available. We introduce HyDRA (Hybrid Denoising Regularization Adaptation), a measurement-only framework for DEQ training that combines measurement consistency with an adaptive denoising regularization term, together with a data-driven early stopping criterion. Experiments on sparse-view CT demonstrate competitive reconstruction quality and fast inference.

Urban traffic optimization must simultaneously balance conflicting objectives such as travel delay, congestion, and emissions, rendering the problem highly nonlinear, dynamic, and multimodal. Existing optimization and learning-based approaches frequently struggle in these conditions, either converging prematurely or failing to preserve diverse Pareto-optimal behaviors. This study introduces EvoLLM-D, a hybrid evolutionary framework that couples Large Language Model (LLM)-guided semantic niching with the Multiobjective Evolutionary Algorithm Based on Decomposition (MOEA/D). The LLM provides context-aware variation proposals and adaptive niche allocation, enabling exploration of multiple meaningful operating regimes, while MOEA/D ensures stable convergence across decomposed subproblems. This integration mitigates sensitivity to manual niching parameters and enhances robustness in high-dimensional landscapes. EvoLLM-D is evaluated on the DLR Urban Traffic (DLR-UT) dataset, a multimodal Unmanned Aerial Vehicle (UAV)-captured real-world dataset, as well as standard multimodal multi-objective benchmarks. Across both settings, EvoLLM-D delivers consistent gains in hypervolume, solution diversity, and Pareto-front stability over NSGA-II, MOPSO, MOEA/D, and the recent LLM4MOEA baseline. The results highlight EvoLLM-D as a scalable, practical, and interpretable solution for multi-objective urban traffic control and, more broadly, for complex real-world optimization problems.