Browse Articles

ViT-MultiRAGNet: A scalable and reliable retrieval-augmented Vision Transformer framework for memory-guided feature fusion multi-modal mammogram classification

Background Diagnosing breast cancer with mammography continues to be an incidentally challenging process because of the numerous different ways of acquiring the images, the vast difference in appearance between lesions, and the minimal use of historical information regarding patients being considered when making a diagnosis. In addition, many current deep learning algorithms use static weight values based on the most recent image taken only, often failing to relate a current case to similar historical examples and to incorporate retrieval-based contextual evidence during inference. Introduction ViT-MultiRAGNet is a retrieval-augmented Vision Transformer framework designed to integrate multi-view 2D mammographic features with memory-bank-based contextual evidence from historically similar cases. The proposed approach enhances diagnostic accuracy by combining global 2D mammographic features, semantic lesion representations, and historical case evidence retrieved from a fold-wise memory bank during inference. Methods Multi-view 2D mammograms were processed using a Vision Transformer encoder to extract global contextual features, while a retrieval-augmented memory bank was used to identify diagnostically similar historical cases for evidence-guided feature enrichment. A Retrieval-Augmented Generation (RAG) module retrieves the top-k most similar feature embeddings from a fold-wise training memory bank and combines them with the current query representation using multi-head cross-attention. The evaluation of the models included utilizing accuracy, area under the receiver operating curve (AUC-ROC), Dice Coefficient, and F1 score metrics at a 95% bootstrapped confidence interval using five-fold stratified cross-validation, with findings assessed using Wilcoxon signed-rank statistical tests. Results The accuracy of ViT-MultiRAGNet is calculated as a value of 0.978 ± 0.009 and AUC-ROC 0.998 ± 0.009 (Wilcoxon p < 0.01), which is significantly higher than unimodal benchmarks and fusion techniques on the RTM dataset. The calculations for the CBIS-DDSM evaluation were 0.961 ± 0.011 for accuracy and 0.989 ± 0.010 for AUC-ROC. The calculations for segmentation evaluation were 0.882 (RTM) and 0.795 (CBIS-DDSM), however both have a higher degree of alignment at the edge. The RAG mechanism improved retrieval-guided fusion compared with simple feature concatenation while maintaining efficient inference performance of 0.31 seconds per image. Conclusion The combination of ViT-based global 2D mammographic representation, retrieval-augmented fusion, and memory-guided contextual evidence improves diagnostic performance, robustness, and clinical interpretability. This Framework appears to be a good candidate for an Evidence-based Decision Support System which will provide a transparent method of helping clinicians make decisions regarding screening and diagnosis of breast cancer.

Optimization of TiO2, SiO2, and ATH filler loadings in silicone rubber composites using Box-Behnken response surface methodology

Abstract The electrical industry is continually evolving to develop advanced materials for reliable insulation of wires and components in diverse environments. Silicone, recognized for its inherent flexibility and high thermal resistance is a predominant material in this field. The incorporation of various fillers further enhances its performance for specific applications. In this study, aluminum trihydroxide (ATH), titanium dioxide (TiO 2 ), and silicon dioxide (SiO 2 ) were employed as commonly used fillers in silicone insulating composites. Critical processing parameters, including the concentrations of ATH, TiO 2 , and SiO 2 , were reviewed using Design Expert software via a Box-Behnken experimental design. Each parameter was examined at three levels: ATH (10, 15, and 20 phr), SiO 2 (20, 30, and 40 phr), and TiO 2 (20, 30, and 40 phr), selected based on the characteristics of the base silicone polymer and the fillers. The properties of the prepared composites were assessed through tensile tests, contact angle measurements, dielectric strength, hardness tests, and rheological analyses. The experimental results demonstrate that the type and content of fillers significantly influence the material’s properties. All formulations met the minimum dielectric strength threshold of 50 kV/mm, confirming electrical suitability. Optimization therefore focused on mechanical and surface properties. It was observed that the highest tensile strength (approximately 4.28 MPa), hardness (around 81 Shore A), and tensile modulus (about 8.67 MPa) were achieved at higher filler loadings. The results indicate that filler type and loading level significantly influenced all measured properties, with optimal formulations identified via regression modelling.

Maternal knowledge of obstetric danger signs and related factors among antenatal care attendees in Hargeisa, Somaliland: A cross-sectional analysis

Background Maternal mortality remains a major global health challenge, disproportionately affecting low- and middle-income countries. Despite international efforts, nearly 95% of maternal deaths occur in these regions, with Sub-Saharan Africa bearing the highest burden. Limited knowledge of obstetric danger signs is a key factor contributing to delays in seeking care, which in turn increases the risk of preventable maternal complications. Evidence on maternal awareness in Somaliland is scarce. Objective This study aimed to assess the level of knowledge of obstetric danger signs and associated factors among pregnant women attending antenatal care in Hargeisa, Somaliland. Methods A facility-based cross-sectional study was conducted from July 2 to August 4, 2022, among pregnant women attending antenatal care services. A total of 222 participants were selected using a systematic random sampling technique from eight maternal and child health centers. Data were collected through face-to-face interviews using a pretested and structured questionnaire and analyzed with SPSS version 25. Bivariate and multivariate logistic regression was carried out. Results About 35.6% had a good knowledge specifically about pregnancy related obstetric danger signs. Only 57 women (25.7%) demonstrated overall good knowledge of obstetric danger signs across pregnancy, childbirth, and the postpartum periods. Knowledge was highest for vaginal bleeding (65% − 67%), while less visible complications such as convulsions and blurred vision were less frequently mentioned. Multivariate analysis revealed that secondary level education or higher (AOR = 3.6, 95% CI: 1.8–7.3), access to and usage of media (AOR = 2.6, 95% CI: 1.2–5.6), and institutional delivery (AOR = 3.4, 95% CI: 1.6–7.2), were significant predictors of knowledge. Conclusion Maternal knowledge of obstetric danger signs in Hargeisa was low. Targeted health education, broader use of mass media, and promoting institutional delivery are critical strategies to improve and reduce preventable maternal mortality.

Geovisualizing land degradation risk in Southeast Brazil Using remote sensing and GIS-based assessment

Abstract Land degradation poses a critical threat to ecosystem services and sustainable development, especially in regions experiencing rapid land-use transitions. This study employs an integrated geospatial approach—combining remote sensing, geographic information systems (GIS), and spatial multi-criteria analysis—to assess and visualize land degradation risk in a strategic coastal region of southeastern Brazil (Rio de Janeiro State). Using the United Nations Environment Programme’s Priority Actions Programme Regional Activity Centre (UNEP-PAP/RAC) framework applied to recent satellite imagery, we generated spatially explicit maps classifying land into stable and unstable categories. A geospatial prioritization model incorporating biophysical and socio-economic variables was developed to identify conservation hotspots and support decision-making. Results show that 68.4% of the landscape is stable, largely consisting of unmanaged areas with agricultural and forest potential, while 7.8% is unstable, with sheet erosion concentrated at agricultural frontiers. Priority mapping classified 51.7% of the area as Stable Medium Priority, revealing widespread latent vulnerability, and 4.6% as Unstable High Priority, necessitating urgent intervention. Complementary analysis of land-use change (1985–2024) highlighted a 199% urban expansion and a 34% decline in agricultural mosaics, underscoring anthropogenic drivers of degradation. This study not only validates the PAP/RAC framework in a humid tropical coastal setting but also delivers actionable geovisualization outputs and a spatial decision-support tool for targeted land management, contributing to soil conservation and sustainable development policy in Brazil.

Optimization of injection molding parameters for bending and tensile strength of ABS and ABS/carbon fiber composites using Taguchi–Grey correlation analysis

This study investigates the effects of injection molding parameters on the flexural and tensile strengths of neat acrylonitrile–butadiene–styrene (ABS) and woven carbon fiber fabric–reinforced ABS composites (ABS/CF) fabricated by a two-step insert injection molding process. A Taguchi L25 orthogonal experimental design was employed with five processing parameters: filling pressure (104–118 MPa), packing pressure (96–100 MPa), filling time (1–3 s), packing time (0.5–2.5 s), and melt temperature (228–232 °C). The results show that flexural strength increased from 5.86–6.49 MPa (ABS) to 5.76–6.93 MPa (ABS/CF), while tensile strength increased from 33.35–35.17 MPa to 38.20–41.27 MPa, corresponding to improvements of approximately 7% and 17–18%, respectively. ANOVA results indicate that filling pressure is the most significant factor, with p-values of 0.012 and 0.039 for flexural performance and a maximum contribution of 47.08% for tensile strength. The optimal parameter combination identified by Taguchi–Grey analysis yields Grey relational grades of 0.725 and 0.728 for flexural and tensile strength, respectively. Confirmation experiments show good agreement with predicted values, with deviations of 1.59% (flexural) and 2.98% (tensile). Overall, this study provides a comprehensive understanding of the process–structure–property relationships in injection-molded ABS/CF composites and demonstrates that the Taguchi–Grey approach is an effective method for simultaneous optimization of multiple mechanical properties. The findings offer practical guidelines for improving the mechanical performance of fiber-reinforced thermoplastic composites in thin-walled and structural applications.

Quantifying the added value of foot-controlled force variables in predicting mild cognitive impairment

Abstract Predicting early cognitive decline is important for timely interventions. However, existing approaches for early detection remain limited by cost, scalability, or concerns about their validity. This study evaluated the potential of lower-extremity force control capacities, a largely unexplored area, to improve prediction of cognitive status in older adults with and without Mild Cognitive Impairment (MCI). MCI refers to an early stage of cognitive decline that does not meet criteria for dementia. 224 participants aged 80–91 years completed a foot-pedal tracking task in which they continuously adjusted the force applied to a pedal to match a visual target curve presented at two different frequencies. The task was designed to capture ongoing visuomotor monitoring and fine motor control. Task performance was quantified using measures of tracking accuracy (Root Mean Square Error, RMSE) and signal complexity (Sample Entropy, SampEn). These reflect how precisely participants controlled force and how regular their output was over time. Participants with MCI showed lower accuracy and lower complexity than cognitively healthy individuals. Adding RMSE and SampEn to a baseline prediction model improved model performance, with the largest gains observed at the higher curve frequency: RMSE contributed 13% new predictive information, SampEn 17%, and both variables together 21%. These gains indicate that foot-controlled force measures added meaningful information beyond established predictors, although confirmation in independent samples is needed. Foot-controlled force tasks may therefore provide a quick-to-administer, objective complement for early detection of cognitive decline.

Interoceptive autonomic regulation in typical development and autism spectrum disorder: A computational model integrating multiple physiological systems

Background Interoceptive cardiovascular signals, including heart rate (HR) and blood pressure (BP), arise from coordinated sympathetic (SNS) and parasympathetic (PSNS) regulation and contribute to affective and cognitive processes. Although atypical autonomic nervous system (ANS) modulation has been reported in autism spectrum disorder (ASD), the dynamical structure underlying branch-specific coordination remains insufficiently characterized. Objective To estimate latent ANS regulatory structure in typically developing (TD) and ASD individuals using a computational modeling framework. Methods A closed-loop computational model integrating cardiovascular, respiratory, and autonomic dynamics was developed. ANS regulation was formalized using three autonomic control modes (coupled reciprocal, coupled nonreciprocal, and uncoupled) and parameterized by effective modulation weights of SNS and PSNS. HR and BP responses to the head-up tilt (HUT) test were simulated, and regulatory surfaces were compared with empirical HR and BP data from TD and ASD groups. Additional simulations under normal respiration, deep respiration, and absence of respiration evaluated mean arterial pressure (MAP) regulation across varying SNS–PSNS activity combinations. Results TD individuals exhibited differentiated SNS–PSNS coordination patterns across control modes, whereas ASD individuals showed convergence of SNS–PSNS weight. In TD, HR and BP distributions under coupled reciprocal mode were most consistent with expected physiological responses, characterized by high SNS and low PSNS activity during postural challenge. In ASD, empirical data extended toward regions associated with relatively higher PSNS weighting, suggestive of persistent parasympathetic engagement during postural challenge. Incorporation of deep respiration enhanced MAP reduction during BP recovery, particularly under over-elevated SNS activity. Conclusion This study provides a mechanistic, state-space characterization of autonomic coordination in TD and ASD populations, enabling inference of latent autonomic regulation from measurable interoceptive phenotypes and identifying respiration as a model-based regulatory lever that augments cardiovascular stabilization.

Shear strength and damage model of silty sand in Daxing 'anling under freeze–thaw cycles

Abstract This study examines the evolution of mechanical properties of silty sand from the Daxing’ anling region under coupled freeze–thaw cycles and confining pressure, using triaxial testing. The results show that at low confining pressure and with zero or limited freeze–thaw cycles, the stress–strain response exhibits strain softening, indicative of brittle failure. In contrast, at high confining pressure or after multiple freeze–thaw cycles, the behaviour transitions to strain hardening, characterised by plastic failure. Freeze–thaw cycling markedly degrades the soil’s mechanical performance. The first cycle causes a 30–40% reduction in peak deviator stress, elastic modulus, and internal friction angle. Subsequent cycles lead to progressively slower degradation, the degradation rate decreased and approached stabilization within the investigated range of 30 cycles. Although increasing confining pressure enhances mechanical behaviour by promoting soil densification, this benefit diminishes as the number of freeze–thaw cycles increases. Based on experimental data, a damage model incorporating the elastic deformation phase was developed. By linking Weibull distribution parameters to test variables, a coupled freeze–thaw and loading damage variable was formulated. Model validation shows close agreement between predicted and experimental results, accurately capturing the effect of freeze–thaw cycles on strength degradation and damage evolution during loading. These findings provide a reliable theoretical basis for the safety design of engineering structures in cold regions.

Examining the feasibility and preliminary effects of resistance exercise training and creatine supplementation in individuals treated for colorectal cancer

Purpose The purpose of this trial was to assess the feasibility, acceptability and safety of a 10-week hybrid (in-clinic and virtual) resistance exercise training RET program with or without CrM in individuals treated for colorectal cancer. Methods Twenty-seven participants were randomized to RET plus 5 grams/day of CrM (EXSUPP; n = 13) or 5 grams/day corn-starch maltodextrin placebo (EXPLA; n = 14). RET was performed three times per week. Feasibility was assessed through recruitment, retention and fidelity (percentage of prescribed RET and supplementation completed). Acceptability was evaluated using a 5-point Likert scale, and safety was monitored through adverse event reporting. Secondary outcomes (body composition, muscular strength, physical function) were assessed pre- and post-intervention using baseline-adjusted ANCOVA models and focus group interviews. Results Registry-based recruitment identified 1378 potentially eligible individuals and 27 of 410 assessed (6.6%) enrolled, highlighting recruitment challenges. Retention was high (24/27; 88.9%). Adherence to RET and supplementation was strong (both > 85%) with no serious adverse events reported. Participants reported high acceptability for both in person and virtual components. No significant between group differences were observed for secondary outcomes; however, both groups demonstrated modest improvements in muscular strength and short physical performance battery (SPPB) scores. Conclusions A hybrid RET program with CrM was feasible, acceptable and well tolerated in individuals with colorectal cancer who were previously treated with chemotherapy. Implications for Cancer Survivors Hybrid supervised RET is safe, acceptable, and associated with modest improvements in strength and physical function among individuals treated for colorectal cancer. While creatine supplementation did not demonstrate clear additive effects in this pilot trial, further adequately powered studies are warranted. Trial Registration: NCT06420726.

A deep neural network for automatic prediction of mouse behavior from ultrasonic vocalizations emitted during social interaction test

Impact of blood pressure control during pregnancy on long-term maternal cardiovascular health – secondary analysis of a follow-up study 15 years post preeclampsia (PAVA study)

Background Within the PAVA study, we investigated placenta-associated vascular aging in women 15 years after pregnancies complicated by preeclampsia and/or fetal growth restriction. During these pregnancies, blood pressure values >140/90 mmHg were considered to require treatment. Following evidence demonstrating improved pregnancy outcomes with tighter blood pressure control, updated German and international guidelines (International Society for the Study of Hypertension in Pregnancy, ISSHP) now recommend target values below 135/85 mmHg. To evaluate whether blood pressure levels during pregnancy are associated with long-term maternal cardiovascular health, we analyzed the impact of blood pressure control during pregnancy on cardiovascular risk profiles obtained in the PAVA follow-up study. Methods Between August 2019 and December 2022, 53 women were examined an average of 15 years after experiencing preeclampsia and/or fetal growth restriction. At the study visit, baseline data were collected, and a comprehensive clinical and functional assessment was performed. This analysis compared 35 women who were hypertensive at follow-up (blood pressure ≥140/90 mmHg and/or antihypertensive medication) with 18 normotensive participants (blood pressure <140/90 mmHg and no antihypertensive medication). Data on blood pressure levels during pregnancy were collected retrospectively from medical records. Results Blood pressure data during pregnancy were available for the second trimester in 15 women who later developed hypertension and 4 who remained normotensive, and for the third trimester in 19 and 14 participants, respectively. Data from the first trimester were not available. In group comparisons, systolic blood pressure in the second trimester and diastolic blood pressure in the third trimester were significantly higher in women who were hypertensive at follow-up (155 vs. 140 mmHg, p = 0.027; 96 vs. 87 mmHg, p = 0.026). In adjusted analyses, no statistically significant associations were observed; however, effect estimates suggested a potential association between blood pressure control during pregnancy and lower odds of later arterial hypertension (OR 0.22, 95% CI 0.03–1.53) and concentric remodeling (OR 0.37, 95% CI 0.03–4.29). Conclusions These findings suggest that blood pressure levels during pregnancy may be associated with long-term maternal cardiovascular health. However, given the limited sample size and exploratory nature of the analysis, these results should be interpreted with caution and require confirmation in larger prospective studies.

Diagnostic utility of p57Kip2, p53, and maspin in classifying hydatidiform moles

Mesothelial cells derived extracellular vesicles promote angiogenesis through the transfer of angiopoietin-2

Background Peritoneal mesothelial cells play a critical role in shaping the peritoneal tumor microenvironment and are increasingly recognized as active regulators of angiogenesis in cancers that metastasize to the peritoneum, including gastrointestinal (GI) and ovarian malignancies. Through complex interactions with tumor and stromal cells, peritoneal mesothelial cells contribute to the establishment of a pre-metastatic niche in the peritoneal cavity. Extracellular vesicles (EVs), nanosized vesicles secreted by most cell types, mediate intercellular communication by transferring bioactive molecules such as proteins, lipids, and nucleic acids. While tumor-derived EVs have been extensively studied in cancer progression, the role of mesothelial cells-derived EVs in regulating endothelial function and angiogenesis remains largely unexplored. Methods EVs were isolated from conditioned media of mesothelial cells and characterized before functional assays. Their effects on endothelial cell behavior were assessed using proliferation, migration, and invasion assays, as well as Matrigel-based tube formation and in vivo angiogenesis plug models. A proteome profiler angiogenesis array was used to identify enriched pro-angiogenic mediators. Mechanistic studies were conducted using lentiviral knockdown and overexpression strategies. Results Endothelial cells efficiently internalized mesothelial cells-derived EVs, leading to significant increases in proliferation, migration, invasion, and angiogenesis in vitro and in vivo. Proteomic profiling identified 43 angiogenic regulators within mesothelial cells-derived EVs, with angiopoietin-2 (ANG2) emerging as a key effector. Functional analyses demonstrated that mesothelial cells-derived EVs-induced angiogenesis was mediated through ANG2-TIE2 signaling, with subsequent activation of PI3K, Akt, and ERK1/2 pathways. Importantly, inhibition of ANG2 markedly reduced these angiogenic effects. Conclusions This work establishes that EVs secreted by mesothelial cells promote angiogenesis by reprogramming endothelial cells through ANG2-dependent signaling. These findings uncover a novel mechanism of mesothelial-endothelial communication and highlight the ANG2 pathway as a promising therapeutic target in advanced GI cancers with peritoneal metastasis.

Acute ischemic stroke, active malignancy and non-bacterial thrombotic endocarditis: an exploratory prospective observational study

Abstract About 3–10% of patients with acute ischemic stroke (AIS) have active cancer. Malignancy-associated hypercoagulability (MAH) is an established cause of cancer-associated stroke. Nonbacterial thrombotic endocarditis (NBTE) is a severe manifestation of MAH. However, no prospective study has examined its prevalence and anticoagulation management in AIS. We conducted a prospective observational study at a tertiary center including patients with AIS or transient ischemic attack (TIA) and active malignancy. All patients underwent transthoracic echocardiography, followed by transesophageal echocardiography when indicated. Among 3,491 screened patients, 16 with active cancer and AIS/TIA were included. Eleven (68.8%) showed embolic patterns suggestive of MAH, and 6 (37.5%) had NBTE. NBTE patients numerically more frequently had multiterritory embolic infarctions (100% vs. 60.0%, p = 0.102) and prior ischemic stroke (66.7% vs. 10.0%, p = 0.028). 4/6 were on direct oral anticoagulants (DOACs), while none were on low-molecularweight heparin (LMWH). Under LMWH, vegetations resolved or regressed in 3/4, whereas under DOACs, progression or recurrent embolism occurred in several patients. Twelve-month mortality was high in both MAH (90.9%) and NBTE (83.3%) groups. NBTE was frequently identified in this prospective cohort of patients with cancer-associated stroke and consistently associated with multi-territory embolic infarction. Recognition may enable earlier diagnosis. Larger studies are needed to define optimal anticoagulation strategies.

Comprehensive evaluation of an emergency monovalent SAT1 foot-and-mouth-disease vaccine: Antigen quality, product consistency, and protective efficacy

Foot-and-mouth disease (FMD) remains endemic in Egypt, with periodic incursions of antigenically distinct virus lineages. In 2025, FMDV serotype SAT1/topotype I was detected in Egypt for the first time, creating an immunity gap in a livestock population previously vaccinated against serotypes O, A, and SAT2. This study aimed to evaluate the quality, immunogenicity, antigenic relationships, and protective efficacy of an emergency inactivated monovalent SAT1 vaccine and to assess potential cross-reactivity with SAT2 vaccine strains. An inactivated SAT1 monovalent vaccine was produced under emergency conditions and subjected to in-process and final product quality control testing, including sterility, safety, RT-qPCR verification, and 146S antigen quantification. Neutralizing antibody responses were measured by virus neutralization test (VNT) at 28 days post-vaccination. Antigenic relationships between SAT1 and SAT2 strains were assessed using r₁-value analysis. Protective efficacy was evaluated by homologous challenge in vaccinated calves. All vaccine batches met quality control criteria, with consistent viral titers and 146S antigen content. Vaccinated calves developed a mean homologous VNT titer of 2.175 log₁₀ TCID₅₀, exceeding the WOAH protective threshold (≥1.65 log₁₀ TCID₅₀). Following homologous challenge, 100% (8/8) of vaccinated calves were clinically protected. In contrast, heterologous VNT titers against SAT2 strains were low (0.26–0.30 log₁₀ TCID₅₀), and r₁-values ranged from 0.12 to 0.14, below the WOAH threshold for antigenic matching (0.3). The emergency SAT1 monovalent vaccine induced strong homologous immunity and complete protection against SAT1 challenge. Limited antigenic and serological cross-reactivity with SAT2 confirms the serotype-specific nature of FMD immunity and supports the need for dedicated SAT1 vaccination and ongoing antigenic surveillance.

Physics‑decomposed residual learning with PolyRF‑boost for cold gas thrust prediction

Abstract In this paper, a method of compilation and leveling is provided to accurately predict the average thrust of cold gas propulsion. The proposed method, which is described as PolyRF-Boost and implemented in the form of conceptual framework Physics-Decomposed Residual Learning (PDRL), explicitly divides the target function into two independent components: a physical macroscopic process (( x )) that describes ideal behavior and a microscopic chemical deviation (( x )) that corrects reality gaps. Accordingly, the final estimator is defined as the sum total of two layers as $$\hat{\user2{f}}\left( {\varvec{x}} \right) = {\mathbf{\mathcal{M}}}_{{{\varvec{\theta}}_{{{\varvec{phys}}}} }} \left( {\phi \left( {\varvec{x}} \right)} \right) + {\mathbf{\mathcal{G}}}_{{{\varvec{\theta}}_{{{\varvec{resid}}}} }} \left( {\varvec{x}} \right)$$ , in which () the macroscopic layer is based on a Backbone Polynomial Ridge on physically engineered features (such as $${\varvec{I}}_{{{\varvec{sp}}}}^{2}$$ and $${\varvec{P}}_{{\varvec{T}}} /{\varvec{I}}_{{{\varvec{sp}}}}$$ ) and $$\phi \left( {\varvec{x}} \right)$$ is second-degree polynomial conversions of those characteristics, and is a refining residual based on Gradient Boosting, which models the complex chemical deviations. The dataset used was loaded from the Excel file Propellant_Ranking_FP.xlsx ( http://dx.doi.org/10.5281/zenodo.7765215 , version 1.0.0, released March 24, 2023). After sampling 500 rows and preprocessing, the final dataset contained 500 samples with 21 features. Preprocessing included detecting and removing data leakage (removing 2 features: Total_Impulse and Average_Satellite_Mass due to high correlation), removing 7 identifier columns and 11 redundant correlated features. Then, polynomial feature generation, feature selection with Random Forest (top-k selection; in practice top-12) and correction with Gradient Boosting were performed. To increase noise robustness, the residual layer was trained with an absolute value error function to reduce sensitivity to outliers. Evaluation with 5-fold cross-validation and MAE, RMSE and ( R 2) measures shows that PolyRF-Boost/PDRL achieves competitive performance with a test R2 of 0.9899, MAE of 0.000312, and MSE of 0.000006, outperforming baseline models including Polynomial Regression (R2 = 0.9887), Random Forest (R2 = 0.8056), and Gradient Boosting (R2 = 0.7667); actual vs. prediction and error comparison plots confirm this performance. Experiments on real data show that ( R 2) in the test set exceeds 0.98, which demonstrates the model’s ability to learn the complex physicochemical structure of cold gas propulsion systems and its applicability to propellant ranking and aerospace system design. Finally, a theoretical analysis is also presented—including a decomposition error bound theorem—that shows that separating the physical process (Bias reduced via Ridge) and the dense residual learning (Exponential error reduction with boosting) leads to a tighter generalization bound than the integrated approaches.

Research on design of soft gripper by dielectric elastomer actuator based on quadrangular star honeycomb support

This study proposes a soft gripper with dielectric elastomer actuator (DEA) and 3D printing quadrilateral star honeycomb support due to the insufficient structural stiffness when grasping irregular objects. According to the bending requirement of the soft gripper and the electrostrictive properties of the dielectric elastomer (DE), a quadrangular star honeycomb structure with zero Poisson’s ratio is proposed as the supporting structure for the soft gripper to enhance its stiffness. The mechanical properties of the honeycomb structure were analyzed via the finite element method (FEM) to improve its stress and horizontal compression displacement. By discussing different parameters such as the pre-stretching ratio of the dielectric elastomer and the pre-compression displacement of the honeycomb structure, we analyzed their influence on the bending displacement performance of the soft gripper. The performance of the soft grippers was verified through both simulations and experiments. The improvement soft gripper, composed of two-layer of dielectric elastomer with a 300% pre-stretching ratio and a honeycomb structure subjected to 20 mm pre-compression, exhibits a bending displacement of 13 mm on a voltage of 5000 V and can grasp objects weighting 5 g.

Robust optimization and stochastic simulation of a centrifugal blood pump under uncertain conditions

Chronic disease related emergency department presentations and potential for redirection to alternative acute care settings (“FOCUS” study): A nationwide flashmob study

Background Emergency Departments (EDs) face increasing pressure, driven in part by the complex needs of an ageing population. Many internal medicine ED visits may reflect complications of chronic conditions. Aiming to maintain the accessibility of acute care, evaluation whether such presentations can be managed in alternative settings is necessary. Methods We conducted a prospective multicentre flashmob study in 28 Dutch hospitals over a 24-hour period (November 28 th and 29 th , 2024). All adult non-elective ED patients presenting for internal medicine were included. Primary outcomes were the proportion of visits related to chronic conditions and the potential for redirection to an alternative acute care facility. Data on patient characteristics, case complexity, and organisational factors were collected and analysed using descriptive and comparative statistics. Results Among 203 included patients, 45.3% presented with an acute complication of a chronic condition. Of these patients, 39.4% were receiving chronic care from an internist. Nearly one-quarter (24.1%) of all ED visits were considered preventable. Additionally, 39.9% of patients could have been managed in alternative settings such as acute outpatient clinics or directly to acute admission units. Despite the presence of these alternative care facilities in most hospitals, the lack of structural organisation was reported in 58.0% as a barrier to redirecting patients. Conclusion A substantial proportion of internal medicine patients at the ED reflect an acute on chronic care need. While alternative care pathways exist, organisational barriers hinder their use. Improving access to structured outpatient acute care could possibly reduce ED use and improve the efficiency of acute internal medicine services.

Numerical simulation on impact induced concrete fracture by the material point method with an HJC-TCK constitutive model

Abstract The study of the damage and failure process of concrete medium under impact loading is an important research topic in the field of protection engineering. Due to the high peak value and short duration of impact loads, the dynamic response of concrete medium under such loading presents a high degree of complexity. The Material Point Method, as a mesh-free particle-based approach capable of conveniently handling large deformations, has seen rapid development in the field of computational mechanics due to its ease of programming implementation and high accuracy, providing a new approach to solving impact-related problems in concrete medium. Therefore, this research first investigated the fundamental theory and algorithms of the Material Point Method and developed a corresponding simulation program. Secondly, an HJC-TCK constitutive model was proposed and implemented in the simulation program to simulate the damage behavior of concrete medium under impact loading. Finally, the program was applied to the numerical simulation of concrete slabs subjected to impact. The numerical results demonstrate that this method can effectively reflect the crushing damage characteristics of concrete under compression and shear, as well as describe the mechanical behavior of crack propagation and collapse under tension.