Advancing Open Science

The detection of defects in Printed Circuit Boards (PCBs) is a critical yet challenging task in industrial quality control, characterized by the prevalence of small targets and complex backgrounds. While deep learning models like YOLOv5 have shown promise, they often lack the ability to quantify predictive uncertainty, leading to overconfident errors in challenging scenarios—a major source of false alarms and reduced reliability in automated manufacturing inspection lines. From a Bayesian perspective, this overconfidence signifies a failure in probabilistic calibration, which is crucial for trustworthy automated inspection. To address this, we propose MFE-YOLO, a Bayesian-enhanced detection framework built upon YOLOv5 that systematically integrates uncertainty-aware mechanisms to improve both accuracy and operational reliability in real-world settings. First, we construct a multi-background PCB defect dataset with diverse substrate colors and shapes, enhancing the model’s ability to generalize beyond the single-background bias of existing data. Second, we integrate the Convolutional Block Attention Module (CBAM), reinterpreted through a Bayesian lens as a feature-wise uncertainty weighting mechanism, to suppress background interference and amplify salient defect features. Third, we propose a novel FIoU loss function, redesigned within a probabilistic framework to improve bounding box regression accuracy and implicitly capture localization uncertainty, particularly for small defects. Extensive experiments demonstrate that MFE-YOLO achieves state-of-the-art performance, with mAP@0.5 and mAP@0.5:0.95 values of 93.9% and 59.6%, respectively, outperforming existing detectors, including YOLOv8 and EfficientDet. More importantly, the proposed framework yields better-calibrated confidence scores, significantly reducing false alarms and enabling more reliable human-in-the-loop verification. This work provides a deployable, uncertainty-aware solution for high-throughput PCB inspection, advancing toward trustworthy and efficient quality control in modern manufacturing environments.

The ductile shear zones in northern Guangxi provide a crucial window for understanding Paleozoic collisional deformation and the tectonic evolution of the South China Block. The Jiufeng–Gandong ductile shear zone is located in the western part of the Motianling pluton in northern Guangxi. The penetrative mylonitic foliation within the ductile zone dips toward the ESE at angles of 55°–85°. Kinematic analyses indicate that the Jiufeng–Gandong ductile shear zone experienced sinistral thrust shearing. Anisotropy of magnetic susceptibility (AMS) results show that the shear zone generally strikes in an NNE direction, with a length exceeding 30 km and a maximum width of more than 2.5 km. The flattening degree (E value) of the magnetic susceptibility ellipsoid suggests that deformation within the shear zone is dominated by flattening strain, accompanied by a component of extensional strain. Quartz dynamic recrystallization mechanisms and electron backscatter diffraction (EBSD) analyses indicate that the sinistral thrust shearing occurred at deformation temperatures of approximately 350–650 °C. LA–ICP–MS U–Pb dating of zircons from a mafic mylonite yields a crystallization age of 443.0 ± 2.8 Ma. By integrating macro- and microstructural observations, magnetic fabric data, quartz EBSD fabric analyses, regional published geochronological constraints, and hydrothermal zircon U–Pb ages obtained in this study, we propose that the Jiufeng–Gandong ductile shear zone developed during Caledonian thrusting of the Cathaysia Block onto the Yangtze Block from SE to NW. Under collisional compression, the shear zone underwent medium- to high-temperature sinistral thrust shearing accompanied by dominant flattening strain. These results elucidate the geometry, strain characteristics, and tectonic regime of the Jiufeng–Gandong ductile shear zone, providing new insights into the Caledonian tectonic evolution of South China.

Due to their complexity and nonlinearity, metaheuristic algorithms have become the standard in problem solving for problems that cannot be solved by standard computational solutions. However, the global performance of these algorithms is strongly linked to the population structuring and the mechanism of replacing the worst solutions within the population. In this paper, an Adaptive Artificial Hummingbird Algorithm (AAHA), a new version of the basic AHA, is introduced and designed to enhance performance by studying the impacts of different population initialization methods within a broad and continual migration form. For the initialization phase, four methods—the Gaussian chaotic map, the Sinus chaotic map, opposite-based learning (OBL), and diagonal uniform distribution (DUD)—are proposed as an alternative to the random population initialization method. A new strategy is proposed as a replacement for the worst solution in the migration phase. The new strategy uses the best solution as an alternative to the worst solution with simple and effective local search. The proposed strategy stimulates exploitation and exploration when using the best solution and local search, respectively. The proposed AAHA is tested through various benchmark functions with different characteristics under many statistical indices and tests. Additionally, the AAHA results are benchmarked against those of other optimization algorithms to assess their effectiveness. The proposed AAHA outperformed alternatives in terms of both speed and reliability. DUD-based initialization enabled the fastest convergence and optimal solutions. These findings underscore the significance of initialization in metaheuristics and highlight the efficacy of the AAHA for complex continuous optimization problems.

Personality traits are essential to understanding individual differences in values, attitudes, behaviours, and cognitive-emotional reactions to climate change (CC). Prosocial traits (empathy and altruism) and nature relatedness (NR), that is, the subjective sense of connection with the natural world, have been linked both to pro-environmental behaviours (PEB) and to CC-related psychological distress. As these reactions are increasingly common in the context of CC, it is crucial to distinguish their adaptive components from their maladaptive ones, namely, by identifying which psychological predictors most strongly promote PEB, in order to design targeted interventions and communication strategies that effectively foster sustainable action. This study examined whether CC-worry, CC-distress, and CC-impairment mediate the relationships between prosocial traits, NR, and PEB. A community sample of 577 adults (mean age = 32.62 ± 14.71 years; 64.6% women) completed self-report measures of the abovementioned study variables, and a multiple mediation model using structural equation modelling was tested. Prosocial traits and NR were positively associated with CC-related psychological distress and PEB, and CC-worry and CC-distress showed significant mediating roles, whereas CC-impairment did not. The model explained 40% of PEB’s variance. Overall, CC-worry and CC-distress appear to function as adaptive, motivational processes that link positive traits and nature connection to environmental action, while CC-impairment reflects a maladaptive, unconstructive response that may index the more pathological end of climate change-related psychological distress.

Background and Objectives: Dysglycemia is a major determinant of adverse outcomes in COVID-19, yet the separate contributions of poor glycemic control and glycemic variability (GV) remain incompletely defined. We conducted a systematic review and meta-analysis of observational cohort studies (both prospective and retrospective) to quantify the impact of chronic hyperglycemia and glucose instability on disease severity, intensive care requirements, and mortality in patients with COVID-19. Materials and Methods: We searched PubMed, Scopus, and Web of Science from January 2020 to October 2024 for observational cohort studies reporting clinically relevant COVID-19 outcomes stratified by glycemic control or GV. Dysglycemia definitions varied across studies (HbA1c-based chronic hyperglycemia, fasting glucose, or admission/in-hospital hyperglycemia). GV was assessed using metrics including mean amplitude of glycemic excursions (MAGE), standard deviation (SD), coefficient of variation (CV), or maximum daily glucose difference. Twelve studies met inclusion criteria and were included in qualitative synthesis; five studies were eligible for quantitative synthesis of clinical outcomes. Random-effects DerSimonian–Laird models were applied due to anticipated clinical heterogeneity. Heterogeneity was evaluated using Cochran’s Q, τ², and I² statistics. Results: Overall, 12 observational studies (9 prospective and 3 retrospective cohorts; n = 1,008,310 patients) were included. In quantitative analyses of five eligible cohorts, poor glycemic control was associated with a significantly increased risk of severe or critical COVID-19 (pooled RR = 1.75, 95% CI: 1.45–2.11; I² = 29%), ICU admission (RR = 1.54, 95% CI: 1.18–2.01), and mechanical ventilation (RR = 1.72, 95% CI: 1.31–2.26). Three studies evaluating GV demonstrated a strong association with adverse outcomes (pooled RR = 2.07, 95% CI: 1.71–2.50; I² = 0%); this low heterogeneity should be interpreted cautiously given the limited number of studies. GV remained associated with mortality in multivariable models, indicating that glycemic variability is separately associated with mortality as a clinically relevant prognostic risk marker in hospitalized COVID-19 patients. Conclusions: Both chronic hyperglycemia and elevated glycemic variability are each associated with increased risk of severe COVID-19 outcomes. Glycemic variability appeared to be a consistent, low-heterogeneity prognostic marker of mortality, being separately associated with higher death risk in hospitalized COVID-19 patients, highlighting its potential utility as a dynamic metabolic biomarker. Early identification and targeted management of dysglycemia—especially glucose instability—may improve prognosis in hospitalized COVID-19 patients. PROSPERO: CRD420251250718.

Objective: The safety of intravenous thrombolysis (IVT) for acute ischemic stroke (AIS) patients with pituitary neoplasms is unclear. This study aims to assess IVT’s safety and efficacy in this patient population. Methods: We reviewed PubMed, Scopus, EMBASE, and Web of Science through July 2025 for reports of IVT administration in AIS patients with pituitary neoplasia. We also performed a retrospective analysis of the Nationwide Readmissions Database (NRD) from 2016 to 2022 to compare outcomes of IVT versus no IVT for AIS patients with pituitary neoplasia, and outcomes of IVT-treated AIS patients with versus without pituitary neoplasia. Outcomes of interest include post-stroke functional status, intracranial hemorrhage (ICH), mortality, and pituitary apoplexy. Multivariate regression analyses were performed to adjust for confounders. Results: The literature review identified 5 AIS patients with pituitary neoplasia, of whom 3/5 (60%) experienced intracranial hemorrhage and none developed apoplexy. In the nationwide analysis of 1,246,750 AIS patients, 1661 (0.13%) had concomitant pituitary neoplasm. Among these patients, IVT was associated with higher odds of functional independence at discharge (adjusted OR 2.46 [95%CI 1.56–3.87]), without increased risk of ICH or in-hospital death (p > 0.05). No cases of pituitary apoplexy were observed. Outcomes among all IVT-treated AIS patients did not differ between those with and without pituitary neoplasms (all p > 0.05). Interpretation: Only five cases of IVT for AIS patients with pituitary neoplasia were identified, highlighting a striking lack of clinical data. In a large U.S. cohort of AIS patients, IVT was associated with improved hospitalization outcomes without increased risk of ICH or pituitary apoplexy.

Road transport is a significant contributor to greenhouse gas emissions within the European Union, with Poland showing one of the most pronounced increases since 1990. Motivated by gaps in national inventories (absence of vehicle-level mileage, limited fuel/age/spatial breakdowns, and scarce real-world hybrid performance data), we develop a novel vehicle-level integration method that links administrative vehicle records with scraped online car-sale listings via deterministic and probabilistic record linkage, imputes missing mileage, and applies age- and fuel type-adjusted emission multipliers to estimate per-vehicle emissions. The approach produces high-resolution breakdowns by fuel type, vehicle age and spatial units while explicitly accounting for hybrid vehicle behavior. This study introduces a novel methodology and analytical product that can be used to estimate emissions from the entire Polish passenger car fleet and monitor decarbonization progress. Applying this method to the Polish passenger-car fleet yields total 2024 passenger-car CO₂ emissions of ≈41.4 million tonnes (our estimate aligns closely with independent national figures), with diesel, gasoline and LPG/CNG accounting for roughly 38%, 47% and 12% of CO₂, respectively. We find that hybrids and EVs currently cut fleet emissions by ~2.6% (CO₂), but their higher utilization magnifies their effect, while vehicle ageing increases total emissions by ~1.2% per year. These results demonstrate that integrating microdata substantially improves the monitoring of decarbonization progress.

Antibiotic resistance gene (ARG) monitoring in environmental systems increasingly relies on DNA-based molecular approaches; however, the extent to which DNA extraction strategies bias downstream resistome interpretation remains insufficiently understood. This study systematically evaluated the effects of single versus successive DNA extraction on DNA recovery, microbial community composition, and the abundance and diversity of 385 genes related to antibiotic resistance including ARGs and mobile genetic elements (MGEs) across three contrasting matrices: water, sediment, and fish intestinal tissue. Successive extraction markedly increased DNA yield and detection of functional genes in water and sediment, particularly for low-abundance and particle-associated taxa. Enhanced recovery resulted in higher richness and abundance of ARGs and MGEs and strengthened correlations between intI1, ARGs, and bacterial taxa, indicating that single-cycle extraction may underestimate resistome magnitude and potential host associations in complex matrices. Conversely, fish intestinal tissue, used here as a representative biological matrix, showed limited benefit or even reduced gene abundance with repeated extraction, likely due to rapid depletion of extractable nucleic acids and DNA degradation. While successive extraction improves recovery efficiency, the potential inclusion of extracellular or relic DNA suggests caution in interpreting inflated ARG abundance. Overall, our findings demonstrate that DNA extraction is a matrix-dependent methodological driver that can reshape both quantitative outcomes and ecological inference. Matrix-specific optimization and careful protocol selection are therefore essential for improving data comparability and minimizing methodological underestimation in environmental resistome assessments.