This study presents a machine learning framework for predicting pedestrian accident severity using Amman, Jordan's first complete 10-year traffic dataset (2014–2023). Addressing the critical class imbalance where minor injuries predominate (85%), causing standard models to poorly detect severe cases (< 25% recall), we implement cost-sensitive algorithms and specialized undersampling techniques, such as XGBoost with Balancing the Loss Function (XGBLF) and Random Data Partitioning with Voting Rule (RDPVR), which enhanced learning from underrepresented Major/Fatal cases while maintaining data authenticity. Through mixed-type correlation analysis and statistical testing, vehicle speed, road illumination, vehicle type, driver age, and road conditions emerged as the most significant predictive factors. RDPVR achieved a 63% true positive rate for Major/Fatal injuries, a 2.78-fold improvement over standard classifiers, and XGBLF achieved 95%, but this achievement was on account of the accuracy of the minor cases. Comprehensive interpretability analysis (SHAP, LIME, and Permutation Importance) revealed that heavy vehicles, poor lighting, and high-speed driving strongly predict Major/Fatal outcomes. Notably, the analysis demonstrates Jordan's improved safety trajectory, with 2020–2023 showing reduced severe accidents compared to 2014–2016, indicating measurable policy impact. This study delivers the first interpretable, context-sensitive AI framework for Amman/Jordan pedestrian safety, translating technical insights into actionable recommendations for targeted interventions, urban planning, and data-driven enforcement strategies to reduce pedestrian injury severity in high-risk zones.

Unmanned aircraft are increasingly recognized for their potential to enhance healthcare logistics, offering rapid and reliable transport solutions. Among the many envisioned use cases, emergency medical deliveries stand out as particularly promising due to their immediate societal value. This study investigates the potential of drones operating under U-space to support hospital-to-hospital emergency deliveries in Madrid. Using the GEMMA tool, we modeled and simulated operations with two drone types along direct routes between four hospitals, resulting in six hospital pairs. Drone travel times were estimated and compared against road transport times obtained from the Google Routes API, incorporating one week of traffic data to capture daily and weekend variability. The results show substantial advantages of aerial transport, with time savings ranging from 2 to 26 min, equivalent to 35–58% compared to road transport. Drones consistently ensured deliveries within 15 min, outperforming regular cars (39%) and ambulances or motorcycles in highly congested periods. Sensitivity analysis confirms their reliability in scenarios with strict time constraints, especially under 15 min. These findings demonstrate that drones reduce travel times and improve predictability, providing a robust evidence base for policymakers and regulators to advance U-space integration in healthcare logistics.

Deformable medical image registration is a fundamental task in medical image analysis. While deep learning-based methods have demonstrated superior accuracy and efficiency, they often overlook the critical role of regularization in ensuring robustness and anatomical plausibility. We propose DARE (Deformable Adaptive Regularization Estimator), a novel registration framework that dynamically adjusts elastic regularization based on the gradient norm of the deformation field. The method integrates strain and shear energy terms, adaptively modulated to balance deformation stability and flexibility, and includes a folding-prevention mechanism that penalizes regions with negative deformation Jacobian to reduce folding artifacts and encourage topologically plausible mappings. Extensive experiments on the IXI, OASIS, and MUI-P datasets demonstrate that DARE consistently improves registration accuracy while preserving anatomical plausibility. Compared with classical regularizers and state-of-the-art learning-based methods, DARE achieves higher Dice scores and reduces folding artifacts, while maintaining low strain energy and realistic volume changes in key brain structures. These results confirm that the proposed adaptive regularization mechanism delivers robust, accurate, and physiologically consistent deformation fields across clinical scenarios related to brain registration.

This study examines job performance among judo referees through the lens of personality traits during World Judo Tour events from 2018 to 2022. Sixty-three referees completed an online questionnaire including the Big Five Inventory (BFI) and the Conditions for Work Effectiveness Questionnaire (CWEQ-II). Data were analyzed using descriptive statistics, correlation analysis, and structural equation modeling (SEM). The measurement model showed acceptable validity and reliability, confirming the structural model. Support and resources emerged as the most influential factors affecting job satisfaction (JAS) and organizational role satisfaction (ORS). Incorporating refereeing experience at major events into the model indicated only partial model fit. Findings highlight the role of structural empowerment in mitigating job dissatisfaction among referees. Future research with larger samples should further strengthen the understanding of the relationship between personality traits, empowerment, and job performance.

Background: Breast cancer remains the most common cancer in women worldwide. Treatment has evolved into multimodal approaches, with pathologic complete response (pCR) after neoadjuvant chemotherapy (NAC) serving as a key prognostic marker. The aim of this study was to evaluate the value of inflammatory markers in predicting pCR to NAC in breast cancer. Methods: This cross-sectional study of 74 patients with breast cancer who underwent NAC followed by surgery included demographic, tumor, and immune-inflammatory marker data. Receiver operating characteristic curve analysis and the Youden index were used to determine optimal cutoff values. Univariate and multivariate logistic regression assessed associations between markers and pCR, adjusting for tumor stage, human epidermal growth factor receptor 2 (HER2), and estrogen receptor (ER) status. Results: Our multivariate analysis identified the pan-immune-inflammation value (PIV), HER2 status, and ER status as significant independent predictors of pCR. PIV (OR, 4.28; 95% CI, 1.59–16.88) remained significant among inflammatory markers, while the neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), and platelet-to-lymphocyte ratio (PLR) did not. HER2-positive (OR, 7.45; 95% CI, 2.30–24.15) and hormone receptor (HR)–negative (OR, 7.02; 95% CI, 2.63–18.70) statuses were also strongly associated with pCR. Conclusion: PIV is a robust predictor of pCR in patients with breast cancer receiving NAC, offering a comprehensive reflection of the immune-inflammatory state. Incorporating PIV with tumor-specific markers (e.g., receptor status, Ki-67, grade) may enhance treatment stratification. Further validation in diverse cohorts is warranted.

Functional Safety system (software & hardware) development is typically a V-Model process, which is governed by strenuous regulations & norms. This, along with use case specificity, and the scrupulous nature of functional safety creates various bottlenecks across the V-Model, i.e., redundant aspects of functional safety system development. To alleviate these bottlenecks, we introduce two LLM assistants designed to support key V-Model phases. The first assistant, the Digital Safety Assistant (DSA), provides safety engineers with general knowledge of functional safety norms through Retrieval Augmented Generation, thus decreasing norm and application domain adaptation overhead. We benchmark various models and assess the DSA using an official functional safety Certification exam, where the DSA achieves up to 70%, surpassing typical performance levels. A second assistant, the Automated Testing Assistant, developed through Parameter Efficient Fine-tuning to support the V-Model verification phase, is capable of correctly generating and debugging PLC test code with 93% correctness.

The β-catenin destruction complex (BDC) is a central node in WNT/β-catenin signaling, governing embryonic development and adult tissue homeostasis. Although recognized as a prime therapeutic target in colorectal cancer (CRC) for three decades, its dynamic architecture and biochemical complexity have hindered mechanistic understanding. Here, we systematically mapped the sequence-function landscape of the BDC using tiled base editor screens across four endogenous components—CTNNB1, AXIN1, APC, and GSK3B. Validation studies identified ∼150 previously unreported mutations across these genes that affected WNT/β-catenin signaling. In addition to known cancer-associated mutations, we discovered rare gain-of-function and separation-of-function alleles of AXIN1 and CTNNB1 that provide mechanistic insights into complex assembly and regulation. We describe a region in β-catenin that regulates its binding to TCF/LEF transcription factors and demonstrate that the AXIN1–β-catenin interface is critical for controlling signaling flux through the oncogenic BDC. Mechanistic studies revealed that assembly of the oncogenic BDC is scaffolded by its own substrate β-catenin, establishing an autoregulatory mechanism that represents an unexploited vulnerability in cancers harboring common APC truncations. Our comprehensive mutational resource provides a foundation for understanding WNT/β-catenin signaling mechanisms in health and disease, while revealing strategies for therapeutic intervention in WNT-driven cancers.

Extracellular vesicles (EVs) transport biomolecules that could serve as biomarkers for disease diagnosis and monitoring. The clinical utility of EVs derived from cerebrospinal fluid (CSF) in patients with intradural spinal tumors (IST) has not yet been investigated. Here, we obtained EVs from CSF of adult patients with intraspinal ependymoma (n = 9), meningioma (n = 9), hemangioma (n = 4) and schwannian tumors (n = 7), as well as comparison group (‘CG’, normal pressure hydrocephalus, n = 7), by ultrafiltration. CSF-EVs were characterized by electron microscopy and nanoparticle tracking analysis. EV populations according to the presence of tetraspanins (CD9, CD63, CD81) were measured by imaging flow cytometry (IFCM). CD81+ EVs were more prevalent in the comparison group, meningioma, ependymoma WHO grade 2, and hemangioma, whereas CD9+ EVs were predominant in ependymoma grade 1 and Schwannian tumors. CD63+ EVs per milliliter/CSF differed between ependymoma WHO grades 1 and 2 (FC = 24.6, AUC = 90%, p < 0.05). Based on results from a bead-based multiplex profiling, we selected ITGB1, CD44, CD133 and HLA-DR/DQ/DP for further phenotyping in CSF-EVs using IFCM, in combination with each tetraspanin as double-positive subpopulations. Compared to CG, CD44+ EVs were the most relevant population in CSF from IST patients, followed by ITGB1. Notable differences in absolute (EVs/mL CSF) and relative (percentages of CSF-EVs) levels were: CD44+/CD81+ for ependymoma grade 1 (FC = 196.5 and 34.5; p < 0.01) and grade 2 (%FC = 6.1, p < 0.05); CD44+/CD63+ for meningioma (abs. and %FC > 1000, p < 0.05); ITGB1+/CD81+ for hemangioma (%FC = 4.8, p < 0.05); and ITGB1+/CD9+ for schwannian tumors (abs.FC = 19.8, p < 0.01). In conclusion, we identified distinct EV subpopulations in the CSF of IST patients, potentially facilitating tumor classification.

Quercus petraea (Matt.) Liebl. holds significant ecological and economic value, yet phenotypic variation within its natural populations remains only partially explored. The main objectives of the study were: (1) to quantify the phenotypic variation in the analyzed leaf and acorn traits; (2) to determine the proportion of phenotypic variation occurring among and within populations; (3) to examine the pattern of population-level variation and its relationship with bioclimatic factors. Material and methods A total of 13 leaf traits and 10 acorn traits were analyzed in 64 sampled trees from seven natural populations. Descriptive statistics and population differentiation coefficients were calculated. The significance of differences in traits mean among populations and among trees within populations were assessed. Principal component analysis (PCA) was conducted at both the tree and population levels. Correlations between morphological principal components and bioclimatic variables were examined, and regression models were developed to evaluate relationships between significantly correlated morphological traits and bioclimatic factors. Results The coefficients of variation for the 23 traits ranged widely, from 12.08% for cupule width to 36.61% for petiole length, with an average of 20.36%. Nested ANOVA revealed substantial phenotypic variations both within and among populations. The primary source of variation was at the intra-population level, accounting for an average of 68.25%, which exceeded the inter-population variation (30.34%). The analyzed traits were reduced to six principal components, which together explained 88% of the phenotypic variation among trees. A statistically significant correlation between morphological principal components and bioclimatic variables was found for the first component, specifically with temperature related variables. Conclusions The study confirms that Q. petraea in Bosnia and Herzegovina exhibits considerable morphological variability, primarily within populations, and that leaf traits are strongly influenced by climatic gradients. Conservation strategies should prioritize the preservation of population integrity and patterns of local adaptation patterns, particularly in ecologically marginal or historically disturbed sites.

Traditional fault allocation methods often struggle in networks with multiple branching points, leading to inaccurate fault distance estimates. This problem not only undermines network reliability, but also forces distribution system operators to spend extra time and resources tracing faults in the field, which can compromise both worker safety and service continuity. This paper introduces a novel fault allocation method that leverages routinely available network data from electricity distribution companies-namely their existing network models and fault-impedance measurements recorded by IEDs. By integrating detailed network calculations under fault conditions, the proposed approach locates ground fault point even in complex branching network configurations. The simulation results - derived from multiple ground fault tests in a small segment of a real Bosnian distribution system (names omitted) - demonstrate accuracy and computational efficiency of the method, making it a valuable tool for supporting operator workflows, reducing outage durations, and strengthening the resilience of tomorrow’s smart grids.

This paper proposes the use of the Linux kernel’s ftrace framework, particularly the function_graph tracer, to generate informative system-level data for machine learning (ML) applications. Experiments on a real-world encryption detection task demonstrate the efficacy of using the proposed features across several learning algorithms. The learner is subjected to the problem of detecting encryption activities across a large dataset of files, where function call traces and graph-based features are used. Empirical results highlight an outstanding accuracy of $99.28 \%$ on the task at hand, underscoring the efficacy of features derived from the function_graph tracer. The results were further validated using an additional experiment targeting a multi-label classification problem by identifying the running programs based on trace data. This work provides comprehensive methodologies for preprocessing raw trace data and extracting graph-based features, offering significant advancements in applying ML to system behavior analysis, program identification, and anomaly detection. By bridging the gap between system tracing and ML, this paper paves the way for innovative solutions in performance monitoring and security analytics.

We consider a large-scale data center where a fleet of heterogeneous mobile robots and human workers collaborate to handle various installation and maintenance tasks. We focus on the underlying multi-agent task assignment problem which is crucial to optimize the overall system. We formalize the problem as a Markov Decision Process and propose an end-to-end learning approach to solve it. We demonstrate the effectiveness of our approach in simulation with realistic data and in the presence of uncertainty.

ObjectiveTo evaluate the predictive value of LA strain parameters and LASI for AF recurrence following electrical CV, and to compare them to conventional echocardiographic, biochemical, and clinical markers.MethodsIn this prospective, observational pilot study, 31 patients with persistent AF underwent electrical CV and were followed for six months. Echocardiographic evaluation included LA reservoir, conduit, and contractile strain, left atrial stiffness index, left atrial volume index (LAVI), left atrial appendage (LAA) morphology, left ventricular ejection fraction (LVEF), right atrial (RA) area, and right ventricular systolic pressure (RVSP). AF recurrence was assessed at three and six months.ResultsAt three months post-CV, LA reservoir, conduit, and contractile strain values were significantly negatively associated with AF recurrence (p < 0.001), while LASI and E/E' ratios were positively associated (p < 0.001). At six months, only contractile strain retained prognostic significance (p = 0.008). LVEF showed a positive correlation with recurrence at six months (p = 0.003), potentially reflecting the role of diastolic dysfunction.ConclusionLA strain parameters and LASI are valuable tools for predicting AF recurrence after CV, particularly in the early post-procedural period. Contractile strain may serve as a more reliable long-term predictor, emphasizing the importance of longitudinal atrial function assessment in rhythm outcome prediction. However, given the small sample size and single-center design, these results should be considered hypothesis-generating, requiring validation in larger studies.