Concrete seawall panels were tested under impact loading, and the effect of concrete compressive strength (50, 60, and 75 MPa) and reinforcement ratios of glass fiber-reinforced polymer (GFRP) bars (0.33%, 0.50%, and 0.83%) was investigated. The experimental analysis included the comparison of crack, force, deflection, acceleration, GFRP bar strains, and energy absorption behavior. Moreover, an analytical study was conducted to compare the experimental results to the predicted analytical results. The experimental data showed enhanced force and maximum deflection behavior of seawall panels with a higher reinforcement ratio, whereas concrete compressive strength adversely affected the impact behavior of panels because of brittleness becoming dominant, especially after 60 MPa. The strain and energy absorption of GFRP bars was maximum for the panel with 200 mm bar spacing and 60 MPa concrete compressive strength, whereas the maximum post-cracking energy absorption in concrete was the highest for the panel with 200 mm bar spacing and 50 MPa concrete compressive strength. The SDOF equation reliably estimated the impact behavior of the panels with a compressive strength of 60 MPa or lower and a bar spacing of not less than 200 mm, in the case of the panels with a reinforcement ratio of 0.50% or less. The results demonstrate the maximum design configuration of GFRP bars in seawall structures under a 2 m impact. • Influence of concrete compressive strength and reinforcement ratios on the impact response is studied. • Failure behaviour, deflection, force, acceleration, strain, and energy absorption are investigated. • Analytical verification (SDOF) and comparison with experimental results are conducted. • Optimal design configuration of GFRP-reinforced seawalls under 2 m impact is presented.

Striking the balance between speed and compression ratio: A fast bit-grouping algorithm and adaptive compressor selection for scientific data

Effects of compression ratio and EGR ratio on combustion stability and emission characteristics of natural gas/diesel engines with pre-injection strategy

Seismic behaviours of T-shaped steel-concrete-steel sandwich composite walls under different axial compression ratios

In recent years, encoding explicit mesh surfaces into compact neural representations has emerged as a prominent research direction. Compression ratio and representation accuracy present a fundamental trade-off for evaluating such algorithms. Traditional approaches typically decompose the input mesh into two components: a simplified base mesh and a neural displacement field. However, this paradigm faces inherent limitations. First, employing triangles or quadrilaterals as geometric primitives necessitates the explicit storage of vertex connectivity, incurring substantial memory overhead. Second, existing approaches typically treat base mesh generation as a decoupled preprocessing step, failing to fully leverage automatic differentiation frameworks to optimize the distribution of the base mesh. To address these issues, we propose RTF2Mesh, a method that achieves compact representation using only unstructured point clouds with feature vectors and network parameters. At its core, our approach leverages a meshless vertex-normal representation derived from the Restricted Tangent Face (RTF). Furthermore, we employ the Kolmogorov-Arnold Network (KAN) to encode both the displacement information and the normals of the vertex-normal representation. The KAN is chosen for its superior parameter efficiency compared to traditional Multi-Layer Perceptrons (MLPs). These two improvements enable RTF2Mesh to achieve a more compact neural representation while eliminating the need for explicit storage of vertex connectivity. During decoding, surface normals are reconstructed from the input point cloud using the KAN's learned weights to generate a base surface. The KAN-based network then predicts the displacements of the subdivided base surface, producing a high-resolution triangle mesh. Compared to current state-of-the-art (SOTA) methods, RTF2Mesh achieves highly competitive performance at equivalent compression rates.

This study analyzes the impact of key-value (KV) representation compression on explainability in lightweight Transformer-based sentiment analysis. DistilBERT and MiniLM are evaluated on IMDB, SST-2, and TweetEval using a two-stage compression framework. The proposed approach achieves compression ratios of 2.46-3.05 while maintaining high representational similarity (cosine similarity > 0.994). Predictive performance remains stable across all settings, with accuracy variations within 0.02. From an explainability perspective, compression effects vary depending on dataset characteristics. SST-2 demonstrates strong robustness, maintaining explanation preservation above 0.90 with high contrastive consistency (~ 0.89). In contrast, IMDB and TweetEval show moderate reductions in fidelity and stability, although meaningful explanation structures are still preserved (preservation ~ 0.53-0.74). In some cases, fidelity and matching reliability are slightly improved after compression. These results indicate that KV representation compression improves memory efficiency while largely preserving predictive performance and core explanation structures. However, its impact on explainability depends on dataset characteristics such as input length and structural complexity.

To develop an MS-Res-AttU-Net-based deep learning framework for automatic measurement of vertebral compression ratio (VCR) on lumbar magnetic resonance images and to evaluate its value for image-based assessment of lumbar vertebral fractures. This retrospective study included 92 patients with lumbar vertebral fractures who underwent sagittal T2-weighted MRI. An MS-Res-AttU-Net framework was constructed for vertebral segmentation and automatic VCR calculation. The dataset was divided into a training cohort (n=64) and an independent test cohort (n=28). Segmentation performance was assessed using sensitivity, specificity, accuracy, and Dice similarity coefficient. Agreement between automated and manual VCR measurements was evaluated using correlation, intraclass correlation coefficient, and Bland-Altman analysis. An ablation study was further performed to assess the contribution of residual, attention, and multi-scale refinement modules. The final MS-Res-AttU-Net achieved stable segmentation performance and showed close agreement between automated and manual VCR measurements. The ablation study demonstrated progressive improvement in both segmentation quality and downstream VCR estimation, while A qualitative comparison of four lumbar MR cases showed MS-Res-AttU-Net produced the smoothest and most accurate vertebral contours. Automatic VCR measurement on lumbar MR images is feasible and clinically interpretable. The MS-Res-AttU-Net-based framework may provide a rapid and objective quantitative tool for lumbar fracture evaluation.

Abstract The growing demand for sustainable energy and stringent emission regulations necessitate the development of cleaner alternative fuels for diesel engines. This study investigates the performance and emission characteristics of a binary biodiesel blend derived from Garcinia gummi-gutta (GGG) and Garcinia indica (GI) doped with CaO·Al₂O₃ nanoparticles (NPs) in a compression-ignition engine. Biodiesel is produced via microwave-assisted transesterification, with a yield of 98.9%, and is characterized according to ASTM standards. A Central Composite Design (CCD) is employed to examine the effects of blend ratio (10–30%), nanoparticle concentration (60–180 ppm), compression ratio (14–18), and engine load (40–100%) on brake thermal efficiency (BTE), brake specific fuel consumption (BSFC), and emissions (CO, UHC, and NOx). A hybrid multi-response optimization framework integrating the Desirability Function Approach (DFA), the Grey Wolf Optimizer (GWO), and the Starfish Optimization Algorithm (SFOA) is implemented to determine optimal engine conditions. Results indicate that nanoparticle doping enhances combustion efficiency by improving catalytic activity and oxygen availability. The optimal single-objective conditions yield a maximum BTE of 34.8%, minimum BSFC of 0.172 kg/kW·h, and reduced emissions (CO: 0.031 vol.%, UHC: 13.82 ppm, NOx: 92 ppm). Multiobjective optimization yields a composite desirability value of 0.938, and experimental validation confirms its predictive accuracy, with an average absolute deviation of 5.85%. The addition of CaO.Al₂O₃ NPs increases the BTE by 11.86% and reduce BSFC, CO, and UHC by 1.72%, 38.89%, and 26.80%, respectively. However, NOx emissions increase slightly by 7.49%, attributed to improved combustion behaviour resulting from higher oxygen content, air–fuel ratio, and calorific value. The study demonstrates that GGG-GI biodiesel blends doped with CaO·Al₂O₃ nanoparticles, combined with a hybrid statistical-AI optimization framework, offer a viable pathway to enhance diesel engine efficiency and reduce emissions, supporting sustainable biofuel deployment.

Erectile dysfunction (ED) is prevalent after traumatic cervical spinal cord injury (CSCI), yet predictive tools for postoperative recovery remain underdeveloped. This study aimed to develop a clinical prediction model for erectile function recovery in male patients undergoing posterior cervical surgery for incomplete traumatic CSCI (iCSCI). In this retrospective cohort study, 207 male patients (aged 18-60 years) with iCSCI (ASIA grades B-D) who underwent posterior cervical decompression between 2018 and 2023 were included. Erectile function was assessed using the International Index of Erectile Function-5 (IIEF-5) at protocol-defined 3-month and 2-year postoperative follow-up visits. Improvement was defined as an increase of ≥ 1 severity category. Candidate predictors were screened via univariate analysis (P ≤ 0.20) and correlation assessment, followed by forward stepwise logistic regression. Model performance was evaluated using area under the receiver operating characteristic (ROC) curve (AUC) and internally validated via bootstrapping. Median total follow-up was 41.00 months. Overall, 72.5% of patients showed erectile function improvement. The final prediction model included four independent predictors: preoperative ASIA grade (OR for grade D vs. B: 30.519, P < 0.001), injury level (C0-C3 vs. C4-C7; OR: 5.749, P = 0.012), injury to surgery interval (OR per day: 0.858, P = 0.018), and spinal cord compression ratio (OR per 1%: 0.937, P = 0.002). The model demonstrated robust discrimination (AUC: 0.881) and good calibration (Hosmer-Lemeshow P = 0.194). Bootstrap internal validation yielded an optimism-corrected AUC of 0.850. A nomogram was constructed to facilitate individualized risk estimation. In male iCSCI patients, erectile function demonstrates significant potential for recovery following posterior cervical surgery. The validated four-factor model-presented as a clinical nomogram-enables personalized preoperative risk stratification to guide counseling and rehabilitation planning. External validation is required before widespread clinical implementation.

This study aims to experimentally investigate the performance, combustion, emissions, and vibration characteristics of a single-cylinder, four-stroke, water-cooled variable compression ratio (VCR) diesel engine fueled with diesel and Karanja biodiesel blends (B20 and B30). Experiments were conducted under varying engine load, compression ratio (CR), and hot exhaust gas recirculation (EGR-HOT) conditions. Engine vibration was evaluated using root mean square acceleration (RMS Accel), and a novel integrated approach was adopted to correlate vibration behavior with performance and emission characteristics. The results show that vibration increases with engine load but decreases with higher compression ratios, while under EGR conditions it initially decreases and then rises at higher rates. The results show that vibration increases with engine load but decreases with higher compression ratios, while under EGR-HOT conditions it initially decreases and then rises at higher rates. Compared to diesel, RMS acceleration decreased by 4.3 and 8.49% under load variation, 3.01 and 7.18% under CR variation, and 2.66 and 6.75% under EGR-HOT conditions for B20 and B30, respectively. Emission analysis revealed reductions of 13.3% in hydrocarbon (HC), 6.58% in carbon monoxide (CO), and 17.98% in smoke opacity for B30, although nitric oxide (Nox) increased by 15.8% as compared to diesel. Combustion analysis indicated a 3.43% increase in maximum cylinder pressure (CPMax) and a 14.94% decrease in net heat release (NHR). In terms of performance, brake thermal efficiency (BTHE) decreased by 10.74%, while brake-specific fuel consumption (BSFC) increased by 18.52%. Overall, the study demonstrates that biodiesel blends improve emission and vibration characteristics but slightly compromise performance, while the integrated analysis provides deeper insight into combustion behavior and engine condition.

The aim of this study was to identify risk factors for progressive local kyphosis (PLK) following percutaneous vertebroplasty/kyphoplasty (PVP/PKP) in patients with osteoporotic vertebral fracture (OVF) and to develop a predictive model based on the osteoporotic fracture (OF) classification, integrating baseline characteristics, imaging parameters, and surgical factors. This model aims to preoperatively identify high-risk patients for PLK and to optimize perioperative surgical decision-making. This retrospective cohort study included 374 OVF patients who underwent single-level PVP/PKP with a ≥ 2-year follow-up. Patients were randomly divided into a derivation cohort (n = 267) and a validation cohort (n = 107). Multidimensional data, including demographics, bone mineral density, comorbidities, imaging parameters (local kyphotic angle [LKA], vertebral compression ratio, OF classification, and endplate integrity), surgical details (cement volume/distribution), and clinical outcomes (visual analog scale [VAS]/Oswestry Disability Index [ODI]), were collected. PLK was defined as a ≥ 10° increase in the LKA at the final follow-up compared with that on postoperative Day 1. Predictors were identified using random forest, least absolute shrinkage and selection operator (LASSO) regression, and decision tree analyses. A risk scoring system was developed via logistic regression and validated using receiver operating characteristic (ROC) curves, Hosmer-Lemeshow tests, calibration curves, and decision curve analysis (DCA). Among 267 patients in the derivation cohort, 43 (16.1%) developed PLK. Compared with the non-PLK group, the PLK group had significantly greater vertebral height loss and worse final VAS/ODI scores. The key predictors identified included the reclassified OF system (Type I, II, and III), the presence of an intravertebral cleft, the presence of cardiac disease, and the overcorrection of the LKA (> 4.5° intraoperatively). A 9-point risk score was established (AUC = 0.915; cutoff = 2.5), with a sensitivity of 57.8% and specificity of 85.3%. Validation cohort analysis confirmed robust performance (AUC = 0.881). DCA demonstrated superior clinical net benefit within a threshold probability of 10-50%. In this study, a predictive scoring system integrating the OF classification, imaging features, and surgical factors was developed to identify OVF patients at high risk for PLK. A total score > 2.5 serves as a clinical decision-support threshold, suggesting that these patients may benefit from a multidisciplinary evaluation for alternative stabilization strategies or intensified postoperative surveillance, tailored to clinical judgment and individualized patient factors.

Interlanguage development is often assessed through structural counts that only partially capture how learner language is organized probabilistically. This study examines whether information-theoretic indicators are sensitive to proficiency-linked distributional differences in argumentative interlanguage writing. It analyzes 200 de-identified argumentative essays: 143 L2 English essays labeled B1, B2, or C1 and 57 genre-matched L1 English essays used as a reference sample. The primary analysis focuses on four operational indicators: lexical entropy (Hlex), grammatical divergence from an L1 reference distribution using POS trigrams (KLgram), compression ratio (CR), and the positional concentration index (PCI). The broader analytical framework also includes an exploratory phraseological layer operationalized as entropy over contiguous 3-word lexical sequences. Descriptive results suggested increasing phraseological dispersion across proficiency levels; however, because more than 90% of these sequences occurred only once within the fixed 250-token windows, the estimates were treated as supplemental and were excluded from the primary inferential analysis. Welch’s ANOVA and Games-Howell post hoc tests indicated ordered group differences in the observed analytic sample. Hlex and PCI increased across proficiency groups, whereas KLgram and CR decreased, suggesting broader lexical dispersion, closer local grammatical alignment with the L1 reference distribution, greater global structural regularity, and stronger early information packaging. The findings are interpreted as evidence of probabilistic reorganization within this bounded argumentative-writing corpus, not as a universal model of L2 development or as a mathematical validation of the measures. They should be read in light of the study’s genre, sampling, annotation, fixed-window, metadata, and data-sharing constraints.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56853-3.

The rapid development of digital media demands higher quality in pattern design. Simultaneously, significant redundant information hinders the transmission and sharing of these patterns, creating an urgent need for image compression. However, traditional image compression methods struggle to balance efficiency and quality. To address this, a new image compression model for digital media patterns is proposed, based on K-means clustering and Locally Linear Embedding methods. This model creates an efficient compression solution by integrating dynamic clustering parameter selection based on color histograms, multi-dimensional image segmentation of color and texture, and a local linear embedding dimensionality reduction algorithm using dynamic neighborhood selection. The model achieves a compression ratio of 84% and a Peak Signal-to-Noise Ratio of 41dB, with no significant quality difference before and after compression, validating the effectiveness of the improvements made to the basic methods. In practical application experiments, the model’s multi-scale structural similarity reaches 0.71. When processing a large number of patterns, the fastest response time is 189ms, with a minimum memory usage of 16.1M. The shortest processing time on a single-core processor is 0.34s. These experimental results demonstrate that the model balances compression efficiency and quality, offering superior compression performance, good robustness, and the ability to handle various complex tasks and adapt to different application scenarios, meeting the high standards of the digital media industry for image compression.

The tetrahedral-octahedral phase transition in silica glass is of fundamental significance to high-pressure research due to its important role in geoscience and condensed matter physics. However, the formation and role of five-coordinate structural units within silica glass during this transformation process remain unclear. This paper presents a comprehensive study of ab initio molecular dynamics (AIMD) simulations of SiO₂ glass under high-pressure conditions. As the volume compression ratio (V/V₀) decreases to V/V₀ ≤ 0.59, a five-coordinate hexahedral SiO5 unit (denoted [5]Si) emerges suddenly, accounting for approximately 50% of the polyhedral units. This tetrahedral-to-hexahedral transition exhibits characteristics similar to a first-order phase transition. Within the compression ratio range of 0.5 ≤ V/V₀ ≤ 0.59, the [5]Si hexahedral structural unit becomes the dominant polyhedral species. Research has revealed that the [5]Si hexahedral unit plays a pivotal role in forming the [6]Si octahedral unit. Additionally, the tetrahedral-octahedral transition in SiO₂ glass under high pressure involves a tetrahedral-hexahedral-octahedral phase transformation whose mechanism is analogous to that of GeO₂ glass under extreme conditions. .

MotivationEfficient data compression is crucial for reducing storage and transmission costs associated to vast volumes of nanopore raw sequencing data. Surpassing the state-of-the-art compression performance has been challenging, and all recent progress in this direction either incur a computational performance over-cost or resort to lossy compression schemes, which are not always desirable.ResultsIn this article, we present PDZ, a lossless compression algorithm that outperforms VBZ, the current defacto standard, both in compression performance and computational efficiency. In our experimental evaluation, the compression ratio improvement ranges from 0.87% to 2.84% depending on the dataset, the compression speed is 1.09× to 2.25× faster depending on the hardware, and the decompression speed is 1.01× to 1.52× faster depending on the hardware. Compared to EX-ZD, a compression algorithm with similar compression performance, the speedup factor for both compression and decompression goes from approximately to , depending on the hardware.Availability and implementationPDZ is implemented in C++ as a new compression method within the POD5 format. The source code is available as a fork of the open-source NanoporeTech library at https://github.com/Rafael-Cast/Piecewise-Differential-Zstd-Coder-POD5-Demo.

Spinal cord morphometry provides essential biomarkers of neurological health, but clinical interpretations are confounded by inter-subject variability and a lack of normative references across the full human lifespan. We address this gap by generating the first comprehensive lifespan charts for cervical spinal cord morphometry. We leveraged 30 population-based brain MRI datasets, aggregating 78,269 scans from 41,042 individuals (ages 0–100) whose imaging protocols included cervical cord coverage. To overcome contrast variability, we employed a state-of-the-art contrast-agnostic deep learning segmentation method, extracting cross-sectional area (CSA), anteroposterior (AP) and right–left (RL/transverse), and shape indices (compression ratio, eccentricity, and solidity) from C1 to C7. Normative trajectories were modeled using Generalized Additive Models for Location, Scale, and Shape (GAMLSS). The resulting charts reveal distinct non-linear lifespan changes: rapid growth through childhood and adolescence, peak maturation occurring in early-to-mid adulthood (e.g., mid-30s for CSA), followed by gradual decreases. Significant regional variations along the cervical cord and consistent sex differences (males > females for size metrics) were quantified. Spinal cord trajectories showed strong temporal coupling with brain white matter and brainstem volumes, suggesting integrated CNS development and aging. These lifespan charts provide a robust normative framework, enabling age- and sex-specific centile scoring of individual spinal cord morphometry. This resource offers a critical tool for differentiating typical variation from pathological changes, enhancing the clinical utility of spinal cord MRI in studies of development and neurodegeneration.

Towards a universal JPEG lossless recompression foundation model for pathology images: A transformer context modeling approach.

Approximate computing is a means for energy efficient computing by allowing some degree of arithmetic imprecision for error tolerant processing applications such as image compression. This work presents a high-efficiency image compression architecture which exploits the use of hybrid approximate multipliers in the discrete wavelet transform (DWT)-pipeline. The work presents four approximate variants which are realized using hybrid Wallace, Dadda and Baugh-Wooley multipliers which results in significant reduction in delay, power consumption and hardware resource usage. Approximation is only used in the least significant areas of the calculation, and exact calculations are used for the most significant calculation areas to maintain visual accuracy. Experimental results show great improvements with the fourth variant of the approximate DWT. The proposed design shows a PSNR of 35.9 dB, a SSIM of 0.94 and compression ratio of 1.9 when integrated to the wavelet-based compression scheme. There is little deviation between original and reconstructed images based on the error analysis. Hardware performance analysis in terms of area, power, and delay is performed by FPG synthesis targeting Artix-7 platform. Overall, the obtained results show that strategic use of approximation in selected computational stages allows energy-efficient wavelet-based image compression without noticeable loss in the perceptual quality.

Deploying deep learning classifiers on resource-constrained mobile devices requires simultaneously satisfying three competing objectives: predictive accuracy, demographic fairness, and inference efficiency. No prior work has jointly formalized these as a constrained multi-objective optimization problem, benchmarked a comprehensive strategy set under identical conditions, or defined a Deployment-Feasible Zone (DFZ) as the feasibility-constrained Pareto subset. This paper provides all three contributions. We benchmark eleven optimization configurations on a 2,821-image dataset with 24 intersectional demographic subgroups (worst-case imbalance 35.47:1) under hard constraints ([Formula: see text], [Formula: see text] per attribute, [Formula: see text]MB, [Formula: see text]ms on an entry-level SoC). Three findings emerge. (1) The combination of 3D-aware augmentation and Protected Fairness Pruning (C2) is the Pareto knee point: [Formula: see text] (95% CI: 0.906-0.962), [Formula: see text], 6.3MB, 187ms, confirmed in 94.2% of bootstrap resamples. (2) Standard magnitude pruning is strictly Pareto-dominated by fairness-constrained pruning (PFP) at identical compression ratio - a result grounded in the low-magnitude minority-encoding effect. (3) The Adaptive Trilemma Weight Scheduler (ATWS) yields consistent gains of [Formula: see text]pp [Formula: see text] and [Formula: see text]-0.7pp EOD over fixed-weight training, compatible with any fairness-constrained strategy.

Efficient compression of multimodal biosignals is critical for bandwidth-constrained wearable devices. Existing methods compress surface electromyography (sEMG) and accelerometer signals independently, ignoring their inherent coupling where muscle activation drives limb kinematics. We propose a cross-modal alignment framework that explicitly aligns the latent representations of these modalities during training through frame-wise correspondence and temporal-dynamics consistency. Experiments on two Ninapro datasets demonstrate significant improvements over state-of-the-art methods across compression ratios from 20 to 90%, achieving up to 28% higher correlation coefficients compared to traditional methods and consistent 1-3 dB SNR gains over deep learning baselines. Downstream gesture classification experiments further confirm that our method preserves task-relevant features, maintaining over 80% accuracy at CR = 40% while baselines require CR≥50% for comparable performance. Inference-time latent analysis reveals that alignment training reduces cross-modal drift by 59-64%, validating that the learned consistency persists after the alignment module is removed. Notably, our alignment module serves purely as a training-time regularizer and is completely removed during inference, ensuring that the deployed encoder-decoder system incurs zero additional computational overhead compared to standard single-modality autoencoders. Studies on five backbone architectures confirm that the proposed alignment framework is backbone-agnostic and broadly applicable. Constraint-aware projections onto representative embedded platforms confirm feasibility for real-time wearable deployment.