Beato, M, Allen, M, Jamil, M, and Vicens Bordas, J. Three-season analysis of physical performance in professional English football: Influence of match outcome, match location, and players' position during official competitions. J Strength Cond Res 40(7): e692-e702, 2026-This study presents a unique longitudinal analysis of physical performance across 3 consecutive seasons of the same professional English football club, capturing its progression from League One to the Championship and ultimately to the Premier League. Using official match data, the investigation aimed to determine how physical performance metrics are influenced by contextual factors including match outcome, location, and playing position. Players ( n = 49) were categorized into 5 positional groups: center backs, wing backs, center midfielders, attacking midfielders, and strikers. Matches were classified by outcome (win, draw, loss) and location (home, away), with player positions assigned based on in-game roles. Global navigation satellite systems-derived metrics included total distance covered, high-speed running (HSR), sprinting distance, and high-intensity distance (HID), alongside counts of high-intensity accelerations and decelerations. Significant effects were found for season ( p = 0.005), match outcome ( p = 0.031), and position ( p < 0.001) on total distance; for season ( p < 0.001), match outcome ( p = 0.010), and position ( p < 0.001) on HSR; and for season ( p < 0.001), match outcome ( p = 0.024), and position ( p < 0.001) on sprinting distance. High-intensity distance was significantly influenced by season ( p < 0.001), match outcome ( p = 0.009), and position ( p < 0.001). Accelerations ( p = 0.004) and decelerations ( p < 0.001) were significantly affected only by position. Match location showed no significant effect except on sprinting distance. These findings highlight the increasing physical demands across competitive tiers and emphasize the importance of tailoring physical preparation to both league level and positional requirements.

Genotype imputation is a powerful tool for inferring missing genotype data in large-scale genetic studies. Over the last two decades, multiple imputation algorithms have been developed, steadily improving in speed and overall accuracy. However, accurate imputation of rare and infrequent variants remains a challenge, largely because existing methods rely on local haplotype matching within genomic windows and do not fully exploit the extended patterns of haplotype sharing that span entire chromosomes. Here we present Selphi, a new genotype imputation algorithm that combines the Positional Burrows-Wheeler Transform (PBWT) with a multi-stage haplotype selection heuristic operating across entire chromosomes. When compared to state-of-the-art methods Beagle 5.4, IMPUTE5, and Minimac4, Selphi showed higher accuracy on the 1000 Genomes Project and TOPMed datasets, across all super-populations and allele frequencies. Similarly, Selphi achieved higher accuracy than Beagle 5.4 on the UK Biobank dataset, which translated into improved concordance with hc-WGS GWAS summary statistics at known trait-associated loci and more accurate polygenic risk scores (PRS). Selphi outputs standard VCF files with genotype dosages (DS), haplotype-specific allele probabilities (AP1, AP2), and a per-variant dosage R-squared quality score (DR2), enabling direct integration with downstream analytical pipelines including standard post-imputation quality filtering.

Local feature matching plays a critical role in robotic SLAM and visual localization. However, in weakly textured indoor industrial environments, lightweight appearance-based methods often struggle to learn discriminative and stable local features. To address this challenge, this paper proposes GAEFeat, short for Geometry-Aware Efficient Feature, a lightweight vision-geometric feature learning network. To address the scarcity of specialized training data, we integrated robotic arm pose priors with depth information to automatically generate cross-view supervision signals and surface-normal labels. Based on this strategy, we constructed two complementary datasets, including a simulated dataset and a real-world dataset, to support feature learning and evaluation in weakly textured indoor industrial environments. For feature extraction, we design a dual enhancement mechanism consisting of a geometric auxiliary branch and a geometry-aware enhancement (GAE) module. The former guides the network to perceive local surface structures through surface normal supervision, while the latter utilizes a gating mechanism to achieve deep fusion between geometric priors and 2D texture descriptors. Experimental results demonstrate that GAEFeat achieves strong robustness and high inference efficiency in relative pose estimation, homography estimation, and visual localization tasks, with particularly notable advantages in near-field, weakly textured industrial scenes. The framework achieves an inference latency of only 3.9 ms on the NVIDIA Jetson AGX Orin edge platform, demonstrating its real-time capability and practical potential for deployment in edge computing environments.

To improve the accuracy of an equilibrium (EQ) wall-stress model in strongly nonequilibrium (NEQ) turbulent boundary-layer flows, an adaptive strategy is proposed to determine the optimal modeling interface (also known as the exchange location, the matching location, etc.) from several candidates spanning a range of heights. First, with a new NEQ sensor, the flows are instantaneously categorized as quasi-EQ and NEQ types. For quasi-EQ turbulent flows, the interface is placed at 10–20% of the local boundary-layer thickness away from the wall, with at least two solution points underneath to ensure the accuracy of the large-eddy simulation (LES) input. For NEQ turbulent flows with significant pressure gradients and separation, we propose using an optimal interface that minimizes the local NEQ effect within the underlying inner layer, while ensuring that the wall-shear-stress direction remains aligned with that predicted using the specified innermost interface. The local NEQ effect is evaluated by integrating a moving-time-averaged streamwise pressure gradient from the wall to the interface. Two benchmark NEQ turbulent flows over a periodic hill and a smooth ramp are calculated for verification. Our numerical experiments show that the proposed approach is a practical way to improve the simulation accuracy of strongly NEQ turbulent flows with EQ stress-based wall-modeled LES.

MMGA-KAN Net: KAN-based multi-resolution and multi-scale graph attention network for global and local unsupervised stereo matching

Vicens-Bordas, J, Colomar, J, Altarriba-Bartés, A, Yeto-Jiménez, A, Jiménez, A, Carrera-Prat, J, García, F, Peña, J, and Beato, M. Comparison of external load demands across three competitive tiers in Spanish football: A three-season single-club study. J Strength Cond Res XX(X): 000-000, 2026-The purpose of this study was to compare the match external load demands experienced by a football team competing for 3 consecutive seasons across various competitive tiers (fourth, third, and second) of Spanish football. Independent variables included player position, match location, and match outcome. A total of 747 individual official match observations from 47 male players were recorded using 10-Hz global positioning system devices. External load metrics (relative to minutes played) included total distance, high-speed running (HSR > 21 km·hour-1), sprint distance (>24 km·hour-1), high metabolic load distance (HMLD > 25.5 W·kg-1), accelerations (>3 m·second-2), and decelerations (<-3 m·second-2). Linear mixed models (significance level set at p < 0.05) assessed the effects of competitive level, position, location, and result, including interaction effects. Cohen's d was also calculated with 95% confidence interval. Results showed that HSR and HMLD differed across competitive levels, with professional (second tier) matches requiring higher HSR demands (medium effects), and fourth tier greater HMLD demands than third tier (small effect). Positional differences were present in all metrics except accelerations, with wide roles (wingers and wide backs) being exposed to greater demands than central positions (medium to large effects). Match location had limited influence, though total distance was slightly higher at home matches (small effect). Winning was consistently associated with higher physical demands, particularly in the second tier, where players covered more distance at high intensities (HSR and sprinting) than when drawing or losing (small to medium effects). Interaction effects indicated that competition level modulated the influence of player position, match location, and result on physical demands. These findings suggest that professional football imposes higher physical demands (although not for all parameters, e.g., accelerations) than semiprofessional football. Coaches and practitioners should consider competition level and contextual factors when designing training and recovery strategies, particularly for wide-position players and during high-stakes matches.

Stereo radargrammetry using Synthetic Aperture Radar (SAR) images is a powerful technique for all-weather 3D topographic measurements. However, conventional methods based on local template matching often struggle to establish accurate correspondences in mountainous or vegetated areas due to severe SAR-specific geometric modulations. In this paper, we propose a novel high-accuracy stereo radargrammetry framework by introducing RoMa, a robust Transformer-based deep learning model, for dense SAR image matching. Optical pre-trained deep learning models often suffer from a domain gap. To overcome this limitation, we develop an automated pipeline to construct a patch-based SAR image dataset using a reference Digital Surface Model (DSM) and an SAR projection model. By fine-tuning RoMa on this dataset, the model effectively adapts to the complex non-linear deformations of SAR images. Furthermore, unlike conventional methods, our approach establishes correspondences directly on the original slant-range images without requiring ground-range projection, thereby avoiding image quality degradation caused by pixel interpolation. Experimental results using airborne Pi-SAR2 images demonstrate that the fine-tuned RoMa significantly outperforms conventional methods, achieving an 82.86% matching accuracy at a 10-pixel threshold. In the 3D measurement evaluation, the proposed method achieves the lowest elevation mean error (−1.24 m) and the highest inlier ratio (74.1%), proving its effectiveness in generating accurate, dense, and wide-area 3D point clouds even in challenging terrains.

Accurate cascaded channel estimation is crucial for unlocking the full potential of reconfigurable intelligent surface (RIS)-assisted millimeter-wave (mmWave) systems. While compressive sensing reduces pilot overhead, conventional estimators suffer from severe performance degradation due to off-grid leakage induced by the continuous nature of spatial angles. To address this issue, we propose a two-stage channel estimation framework that divides the estimation process into two sequential phases, namely support selection and amplitude recovery. Based on this framework, we design an algorithm termed TS-PO. In the first stage, a preconditioned linear Bregman iteration (PLBI) mechanism is employed to identify the true channel support. Subsequently, the second stage utilizes a localized orthogonal matching pursuit (OMP) refinement to accurately recover the physical channel gains. Simulation results demonstrate the effectiveness of the proposed TS-PO in suppressing off-grid energy leakage. Specifically, it effectively mitigates the estimation error floor, achieves high reconstruction accuracy under stringent pilot overhead constraints, and exhibits strong robustness in dense multipath environments.

ABSTRACT This study examined positional differences, match-to-match variability, and seasonal exposure to peak locomotor demands in professional soccer players. Global positioning system (GPS) data were collected from 24 outfield players across 46 matches (600 player – match observations). Peak high-speed distance (HSD), sprint distance (SD), and acceleration – deceleration density (AD) were quantified using fixed 1-minute worst-case scenarios (WCS), expressed as absolute values and relative to each player’s individual seasonal maximum (%iWCS). Peak HSD and AD differed by position (p < 0.05), with central midfielders exceeding centre-backs by +10.49 m·min−1 for HSD, while strikers demonstrated greater AD (+0.48 AU·min−1). No positional differences were observed for SD (p = 0.357). Within-player variability ranged from 22–28% (HSD), 45–62% (SD), and 23–42% (AD). Mean relative exposure remained below individual maxima across positions (HSD: 63–69%; SD: 41–49%; AD: 45–60% iWCS), with near-maximal exposures (≥85% iWCS) occurring in only 5–20% of observations. Peak locomotor demands were largely independent of match outcome and location. These findings indicate that worst-case scenario demands are position-specific, highly variable, and infrequently reach near-maximal levels during competition. Practitioners should consider both absolute and relative measures when interpreting peak match demands and ensure that training exposes players to the upper range of observed intensities.

Sparse local feature matching has served as the cornerstone of numerous visual geometry tasks and attracted extensive attention. Although significant progress has been made in this area, improving the discriminative power of descriptors remains a key challenge. As far as we know, existing sparse feature matching methods only predict a single descriptor map for keypoints, which might restrict their potential in solving complex scenarios. This issue is particularly pronounced in real-time applications where most methods only learn descriptor maps at a reduced spatial resolution compared to the input image. Consequently, they require interpolating from the low resolution map for obtaining per-keypoint descriptors, which will introduce background contamination and reduce the discriminability of final descriptors. To address these issues, we propose an efficient novel complementary local feature description model. Specifically, the model simultaneously learns two descriptor maps using different loss functions within a single Convolutional Neural Network (CNN). An orthogonal loss is introduced to effectively coordinate the learning of the two branches, aiming to obtain decoupled and complementary descriptors. Extensive experiments across various visual geometry tasks, such as homography estimation, indoor and outdoor pose estimation, as well as visual localization, have demonstrated the superior performance of the proposed method.

ABSTRACT Determinants of success in professional football are well documented, with technical, physical, and situational factors associated with match outcomes. However, these factors are unexplored in English Non-League Football (NLF). This study investigated match-level performance variables and associations with match outcomes and situational factors. Data from a professional NLF team were collected from 33 matches during one season. Technical (pass completion, possession estimated by pass ratio, and shots) and physical (running speed) metrics were compared, and assessed against match outcome, location, and opposition quality. Results revealed the team was technically superior to its opposition, recording significantly higher possession, passing metrics, and total shots (p < 0.05). However, a paradox emerged in pass ratio dominance: possession was significantly higher in matches lost (56.5 ± 6.2%) compared to matches won (48.3 ± 6.1%; p = 0.011, d = 1.34). Additionally, pass completion was higher in draws than wins (p = 0.045, d = 1.30). Match location did not influence performance metrics. Opposition quality did not impact physical performance metrics, though team total shot volume was higher against lower-ranked opposition (p = 0.029, d = -1.45). These findings highlight that for a professional NLF team, technical superiority and possession-based strategies do not guarantee success.

This study quantified performance indicators associated with match outcomes among champion teams from the five major European football leagues during the 2023–2024 season. Ordinal logistic regression with robust standard errors clustered by team was employed, with analyses stratified by match location (home/away) and opponent quality (high/medium/low). Data from 182 matches were sourced from Wyscout® and included offensive indicators (possession, passes, shots, shots on target, expected goals) and defensive indicators (interceptions, fouls, shots conceded, yellow and red cards). Spearman correlations showed that goals scored (q=0.523) and shots on target (q=0.243) were positively associated with match outcomes, whereas goals conceded (q=−0.441) and fouls (q=−0.255) were negatively associated. Ordinal regression revealed context-dependent effects. Offensively, shots on target increased the odds of a better outcome at home (OR = 3.76) and against high-quality opponents (OR = 5.24), while expected goals (xG) was the key predictor in away matches (OR = 2.09). Defensively, interceptions were crucial against high-quality opponents (OR = 1.76), while fouls (OR = 0.53) and yellow cards (OR = 0.61) were detrimental against medium-quality opponents. Against low-quality opponents, shots on target conceded (OR = 0.22) and red cards (OR = 66.58) were critical. Volume-based indicators did not retain significant independent effects. For elite champion teams, competitive success is predominantly determined by efficiency-based indicators, shot accuracy, expected goals, and defensive organisation, whose relevance varies systematically with context. These findings provide exploratory insights and a context-sensitive benchmark for performance analysis at the highest level of European football, warranting further validation in future studies.

TFDM-CR: Time–frequency diffusion modeling for lithium-ion battery capacity prediction incorporating regeneration phenomena

Parallel consensus transformer for local feature matching

Since the failure of the Scan Line Corrector (SLC) on the Landsat 7 ETM+ sensor in 2003, approximately 22% of pixels in acquired images are missing in a regular stripe pattern, severely limiting their application in time-series analysis and land cover monitoring. To address the limitations of traditional methods in modeling complex land surfaces and their tendency to introduce spectral distortions, this paper proposes a stripe-aware deep learning restoration network called SINet (Stripe Inpainting Network). The network leverages a Stripe Attention Module (SAM) to exploit the geometric prior of stripes and aggregate features along the stripe direction, and a Spectral Reconstruction Module (SRM) to model multi-band correlations and preserve spectral fidelity. A hybrid loss function combining pixel loss, perceptual loss, spectral angle loss, and gradient loss is designed. Experiments conducted in the Zhengzhou-Kaifeng-Xuchang junction area of Henan Province show that SINet achieves a PSNR of 38.2 dB, an SSIM of 0.978, and an SAM of 1.82° on simulated data, significantly outperforming baseline methods such as nearest neighbor interpolation, bilinear interpolation, and local histogram matching. When applied to land cover classification, the overall accuracy improves from 74.3% (uncorrected) to 88.5%, approaching the 90.2% accuracy of original Landsat 8 imagery. This study provides an effective solution for restoring historical Landsat 7 data.

Urban microclimates are traditionally interpreted and modelled based on permanent built surfaces, while semi-permanent elements such as stationary vehicles remain largely overlooked in urban climate studies. Despite their distinct radiative and thermal behaviour and increasing spatial prevalence in contemporary cities, parked vehicles are rarely represented in urban microclimate modelling frameworks. This study provides an exploratory assessment of a commonly used three-dimensional modelling workflow to approximate near-vehicle air temperature patterns at the micro-scale by comparing simulated results with field measurements collected in Lisbon, Portugal. Air temperature was measured in an open parking environment with unobstructed sky exposure, at multiple heights above two black and two white parked vehicles during summer, and corresponding simulated values were extracted at matching locations. Simulated mean air temperatures showed reasonable agreement with observations (MAE = 0.6–0.9 °C; RMSE = 0.7–1.3 °C), suggesting that simplified modelling approaches can reproduce general air temperature patterns under controlled conditions. However, larger localised deviations were observed near vehicle surfaces and rear positions, particularly for dark-coloured vehicles, highlighting limitations in resolving fine-scale radiative and aerodynamic processes. The findings indicate that stationary vehicles can be represented as distinct urban surfaces, while emphasising the need for improved parameterisation to enable their integration into urban microclimate models at larger spatial scales.

Ultrasound images have some limitations, such as low signal-to-noise ratio (SNR), speckle noise, lower dynamic range, blurred boundaries, and shadowing; therefore, ultrasound image registration is an important task for estimating tissue motion and analyzing tissue mechanical properties. In this paper, an effective non-rigid ultrasound image registration method is proposed. By integrating intensity, local phase information, and descriptor matching under a variational framework, we can find and track the non-rigid transformation of each pixel under diffeomorphism between the source and target images based on the warping technique. Experiments using simulation and in vivo ultrasound images of the human carotid artery are conducted to demonstrate the advantages of the proposed algorithm, which will act as an important supplement to current ultrasound image registration.

Background: The majority of the funding for the New Rural Cooperative Medical System (NCMS) is derived from fiscal subsidies, comprising central transfer payments and local fiscal matching subsidies. Local governments' strategic behavior in response to central transfer payments may further impact NCMS compensation spending and medical economic risks. Methodology: Accordingly, this paper investigates, from both theoretical and empirical perspectives, the impact pathways through which local fiscal matching subsidies influence the medical economic risks faced by insured rural households, with central transfer payments serving as a moderating factor. This paper constructs a dynamic game framework involving the central government, local governments, and household sectors. It further applies a mediation effect model and related econometric methods to conduct empirical analysis using 87,630 observations from the China Family Panel Studies (CFPS). Results: The results show that, first, local fiscal matching subsidies significantly reduce catastrophic health expenditures for rural households under the income effect of central transfer payments. However, under the substitution effect, the opposite occurs, as local governments adopt non-cooperative strategies in response to central transfer payments. Second, these impacts exhibit regional heterogeneity, with stronger effects in eastern regions, regions with more developed secondary industries, and regions with higher fiscal self-sufficiency rates. Third, local fiscal matching subsidies influence medical economic risks through compensation spending, under the moderating role of central transfer payments. Conclusions: This paper provides a novel perspective on why the NCMS struggles to provide effective protection, thereby enriching the existing literature. Furthermore, it provides policy guidance for fiscal and healthcare reforms in countries with similar contexts to China. Based on these insights, we argue that, during the future integration process of the Basic Medical Insurance for Urban and Rural Residents, clear boundaries should be defined for local fiscal matching subsidies, and the moderating role of central transfer payments should be strategically leveraged.

High-quality, densely annotated data serve as a crucial foundation for developing robust X-ray angiography segmentation models. However, obtaining per-object pixel-level annotations in the medical domain is both expensive and time-consuming, often requiring close collaboration between clinical experts and developers. This paper aims to reduce the annotation costs of X-ray angiography videos by leveraging few-shot video object segmentation (FSVOS), which separates target objects from the background using only a single annotated frame during inference. We introduce a novel FSVOS model that employs a local matching strategy to restrict the search space to the most relevant neighboring pixels. Rather than relying on inefficient standard im2col-like implementations (e.g., spatial convolutions, depthwise convolutions and feature-shifting mechanisms) or hardware-specific CUDA kernels (e.g., deformable and neighborhood attention), which often suffer from limited portability across non-CUDA devices, we reorganize the local sampling process through a direction-based sampling perspective. Specifically, we implement a non-parametric sampling mechanism that enables dynamically varying sampling regions. This approach provides the flexibility to adapt to diverse spatial structures without the computational costs of parametric layers and the need for model retraining. To further enhance feature coherence across frames, we design a supervised spatio-temporal contrastive learning scheme that enforces consistency in feature representations. In addition, we introduce a publicly available benchmark dataset for multi-object segmentation in X-ray angiography videos (MOSXAV), featuring detailed, manually labeled segmentation ground truth. Extensive experiments on the CADICA, XACV, and MOSXAV datasets show that our proposed FSVOS method outperforms current state-of-the-art video segmentation methods in terms of segmentation accuracy and generalization capability (i.e., seen and unseen categories). This work offers enhanced flexibility and potential for a wide range of clinical applications. Code is available at: https://github.com/xilin-x/XRAVOS.

Sánchez-Abselam, O, González-Fernández, FT, Castillo-Rodríguez, A, and Onetti-Onetti, W. External load of professional female soccer players in the competitive microcycle: Influence of playing position and contextual variables. J Strength Cond Res 40(3): e324-e332, 2026-Women's soccer has experienced substantial growth in recent years, accompanied by increased scientific interest in performance-related variables. This study aimed to analyze the external load across competitive microcycles in a professional women's soccer team and to examine the influence of playing position and contextual factors (match location, outcome, and opponent quality) on physical demands. Eighteen professional players (age: 24.5 ± 5.6 years; body mass: 58.8 ± 14.8 kg; height: 165 ± 5.7 cm) from a Spanish second-division team were monitored across 13 microcycles using 10 Hz Global Positioning System devices. Players were categorized into 5 positions: central defenders, external defenders, midfielders, wingers, and forwards. Significant differences were observed across microcycle days ( p < 0.001), with match day (MD) presenting the highest external load values, and MD-1 the lowest. Playing position significantly affected explosive distance, high-speed running, and high metabolic load distance ( p < 0.001), with forwards showing the highest values and central defenders the lowest. In addition, greater physical demands were recorded when competing against lower-ranked opponents. These findings provide relevant insights for physical performance staff, highlighting the importance of adjusting training loads based on both positional profiles and contextual factors to optimize performance and recovery strategies in elite women's soccer.