Advancing Open Science

Surface defect detection of industrial products is a critical component of intelligent manufacturing quality control. However, existing unsupervised anomaly detection methods still struggle to achieve a favorable balance among image-level detection accuracy, pixel-level localization performance, and inference efficiency. To address this issue, this paper proposes a Fusion PatchCore with Lightweight Diffusion Reconstruction (FPC-LDR) framework for industrial surface defect detection, exploiting the complementary properties of discriminative feature matching and generative reconstruction. The proposed method first constructs a memory-bank module based on deep semantic features to identify anomalies via feature matching. Then, a lightweight latent diffusion reconstruction module is developed, in which single-step reconstruction replaces conventional multi-step diffusion processes, reducing computational overhead while maintaining reconstruction quality. Furthermore, an adaptive fusion strategy is introduced to integrate feature distance and reconstruction residuals, improving detection robustness and localization accuracy. Extensive experiments on the MVTec AD dataset demonstrate that FPC-LDR achieves an image-level AUROC of 94.65%, a pixel-level AUROC of 94.54%, and a PRO score of 95.28%. Cross-dataset evaluations on the VisA dataset further verify its generalization capability. The proposed method achieves an effective trade-off among accuracy, localization performance, and inference efficiency, demonstrating its potential for real-time industrial defect inspection.

Walking is the most sustainable and socially inclusive urban travel mode, yet in many rapidly transitioning cities, obstructed and inaccessible sidewalks undermine it, and walkability assessment tools rarely tell municipalities where to act first. This study develops a transparent, low-data protocol that converts a micro-scale sidewalk audit into an ordered set of candidate intervention priorities, using 52 street-unit sides nested within 26 street units in central Blida, Algeria. Twenty audit variables were aggregated into eight obstacle families and normalised to occurrences per 100 m using digitised street-unit lengths (170–1255 m). A two-profile k-means partition, retained as an operational triage device and stable under street-unit bootstrap resampling, separates lower- and higher-obstruction sides. A random forest with SHAP attribution, validated with folds grouped by street unit, provides an internal ranking of the families that separate the profiles: normative accessibility (27.3%), bulky waste (22.6%), and structural defects (15.9%) lead, with threshold-like effects, and the ordering closely matches simple between-profile contrasts (Spearman ρ = 0.98). Normative accessibility leads under all robustness checks, whereas the order of the second-tier families depends on the number of profiles. Physical works and use regulation carry similar shares of attribution (49.4% and 45.4%); excluding accessibility deficits, which record missing provisions rather than occupied space, shifts the lead clearly to use regulation (62.5%). A minimal agent-based model illustrates how local detours can produce a nonlinear loss of pedestrian performance. The result is a reproducible, low-cost workflow that helps resource-constrained municipalities direct limited budgets towards safe, accessible, and inclusive sidewalks, in line with Sustainable Development Goal 11 (targets 11.2 and 11.7).

Generative Artificial Intelligence (GenAI) is reshaping educational practice while raising pedagogical, ethical, and professional challenges. This study examined Spanish teachers’ self-reported GenAI tool use, perceived knowledge, perceptions and attitudes, and training-related experiences and orientations. A quantitative cross-sectional survey was conducted with 838 teachers from primary, secondary, vocational, and higher education, recruited through non-probability convenience sampling. Data were collected using an adapted and extended 47-item questionnaire comprising three scales derived from CAPIAG-P and a nine-item modality-based GenAI Tool Use scale developed for this study. Internal consistency was good to excellent (Cronbach’s α = 0.827–0.930). Descriptive statistics, Holm-adjusted non-parametric comparisons, Dunn–Holm post hoc tests, effect sizes, and hierarchical multiple regression with HC3 robust standard errors were applied. Use was concentrated mainly on text-to-text applications, whereas perceived GenAI knowledge was moderate. Teachers positively evaluated several educational applications but expressed reservations about critical thinking, educational quality, and replacement of the teaching role. They reported limited previous training and strong agreement with predefined pedagogical and ethical professional-development priorities. Perceived GenAI knowledge was most strongly associated with tool use, followed by perceptions and attitudes; training was not independently associated with use. The final model explained 42.0% of the variance. This pattern remained in a sensitivity analysis excluding knowledge items with the clearest behavioral overlap, although explained variance decreased to 37.7%. Holm-adjusted group differences by sex, school ownership, and teaching level were small, while none remained by age or experience. The findings support equitable institutional provision and continuous, pedagogically and ethically grounded professional development.

Pladienolide B (1), a macrocyclic polyketide isolated from Streptomyces platensis Mer-11107, displays potent antiproliferative and antitumor activities through its interaction with the SF3b subunit of the spliceosome. Herein, we report a divergent, stereoselective synthetic route to access both the C1–C8 and C9–C13 fragments of the core moiety of Pladienolide B (1) starting from a common, inexpensive chiral-pool precursor, L-Malic acid. Key stereocenters and structural motifs were established using Sharpless asymmetric epoxidation, regioselective nucleophilic epoxide opening, Mannich reaction for exo-methylene introduction, and a tandem iodination–zinc elimination cascade. This unified precursor approach offers a streamlined tactical route toward the core macrolide framework.

This study presents a DFT protocol for the substitution kinetics of Pt(II) complexes, applied to cisplatin as a case study. We aim to demonstrate that a computed barrier that agrees with experiment does not, by itself, validate the kinetic model, and that further diagnostics are required to assess the reliability of the results. M06-2X/x2c-TZVPall calculations with implicit solvation, explicit treatment of one coordinating water molecule and a set of physical corrections were used to study the aquation, thiolation (with a capped cysteine-thiolate model) and guanine substitution (by 9-methylguanine) pathways. The calculated precomplex-to-transition-state barrier for the first aquation is 2.0 kcal⋅mol⁻¹ below the Eyring barrier inferred from experiment; with the symmetry factor σ = 2 applied, the calculated observed rate constant is 56-fold higher than the experimental value. The formal low-concentration second-order coefficient for thiolation exceeds the reported glutathione rate coefficient by 1.5–1.8 log units, whereas omission of the physical corrections shifts the calculated rate predictions by about 5–8 log units. A direct selectivity descriptor (S_direct) is introduced and shown to be governed by the precomplex term rather than by the activation step, making it vulnerable to precomplex overstabilization by the implicit solvent. Four diagnostics are proposed that should accompany any computed rate constant or selectivity: the sign of the precomplex free energy, the magnitude of K_pre, the proximity to the diffusion limit, and the decomposition of selectivity into precomplex and activation contributions. These provide a systematic basis for assessing the reliability of computed predictions.

This study investigates changes in local groundwater carrying status and regional differences under the background of reductions in total groundwater abstraction, using Jiuquan, Jiayuguan, Zhangye, Jinchang, and Wuwei in the Hexi Corridor as study areas. The analysis was conducted from three dimensions: resource base, development pressure, and groundwater-level–storage change. The groundwater abstraction pressure index and net abstraction excess were jointly used to distinguish relative development intensity from the absolute volume exceeding allowable abstraction. Regional differences were further identified using groundwater-depth changes at 515 monitoring stations and shallow groundwater-storage changes. Residual correlations after controlling for month effects and linear trends, together with robustness tests, were used to examine the associations of precipitation, river runoff, and groundwater abstraction with groundwater depth. Results showed that total groundwater abstraction in the five cities decreased by 23.8% from 2018 to 2025, whereas the abstraction pressure index in key overexploitation areas remained between 1.12 and 2.08 in 2024, indicating that regional abstraction reduction and local overexploitation coexisted. Jinchang had the highest relative abstraction pressure, whereas Wuwei had the largest net abstraction excess, demonstrating that relative development intensity and excess abstraction volume correspond to different management priorities. During the study period, groundwater levels declined at 404 monitoring stations, and the cumulative shallow groundwater-storage change was −69.14 × 10⁸ m³. Storage losses were concentrated mainly in Zhangye and Jiuquan and did not correspond to the regional ranking of development pressure, highlighting differences between current development intensity and cumulative storage change. After month effects and linear trends were controlled, the absolute values of 14 of the 15 correlations decreased, and 11 correlations changed from positive to negative, indicating that common seasonal cycles and long-term trends substantially affected the raw contemporaneous associations. By combining diagnosis of relative intensity and water-volume excess, comparison of groundwater-level and storage changes, and temporal-structure adjustment, this study refines regional identification of groundwater carrying status and provides a basis for spatially differentiated abstraction reduction and coordinated surface-water–groundwater management.

Valorisation of agro-industrial biomass into functional coatings offers a sustainable approach to developing bio-based materials for postharvest applications. Although polysaccharide–lipid coatings are widely studied for their barrier properties, their effects on fruit–packaging friction remain poorly understood. This study recovered pectin-rich polysaccharides (OP) from okra-processing biomass and developed OP–beeswax (BW) composite coatings to modify mango–packaging friction and evaluate selected postharvest quality attributes. Conventional thermal extraction (CTE) and microwave-assisted extraction (MAE) yielded OP with comparable degrees of esterification but distinct hydration properties, with extraction yields of 4.28% and 5.47%, respectively (p > 0.05). The recovered OP exhibited low-methoxyl characteristics, with a degree of esterification of approximately 47%. Film optical and microstructural properties varied with OP and BW levels. Coating performance was evaluated on three independent mango fruits per treatment (n = 3; 21 fruits in total), with the same fruits used for friction and all quality measurements. Static and kinetic coefficients of friction (COFs) at the mango–corrugated plastic board interface were in the ranges of 0.45–0.59 and 0.42–0.58, respectively, demonstrating formulation-dependent frictional behaviour. T3 (1.0% OP + 1.0% BW) exhibited static and kinetic COFs of 0.46 and 0.42, respectively, comparable to the uncoated control (0.46 and 0.45), whereas T4 (2.0% OP + 1.0% BW) showed the highest values (0.59 and 0.58). After three days of storage, coating treatments significantly affected weight loss, firmness, and titratable acidity (TA), with the lowest weight loss observed in T6 (3.25%) and the highest TA in T2 (0.47%), whereas total soluble solids and pH did not differ among treatments. These findings demonstrate the potential of okra-processing biomass for developing bio-based coatings with tunable fruit–packaging friction, although their effectiveness in reducing transportation-induced damage requires further investigation.

Respiratory viruses with broad host ranges are relevant to veterinary epidemiology, biological safety, and One Health surveillance. This retrospective study integrated companion-animal and wildlife records to examine two complementary surveillance questions: SARS-CoV-2 detection in dogs and cats with close human contact, and influenza A virus detection across a broader multispecies panel that included wild birds. We analyzed 256 archived animal samples collected in Kazakhstan during 2022–2023, comprising 158 dogs, 76 cats, and 22 animals of other domestic and wild species. A previously developed in-house multiplex RT-qPCR assay was used to detect SARS-CoV-2 and influenza A virus RNA, and a previously developed in-house SM-antigen-based ELISA was used to assess SARS-CoV-2-associated serum reactivity. SARS-CoV-2 RNA was detected in 104/256 samples (40.6%; 95% CI, 34.8–46.7), including 68 dogs and 36 cats. Influenza A virus RNA was detected in four wild birds (1.6%): two partridges and two mallards. According to the assay-specific classification recorded in the source dataset, 34 sera were reactive, 14 were doubtful, and 208 were non-reactive. All SARS-CoV-2 RT-qPCR-positive samples were ELISA-non-reactive; all reactive or doubtful ELISA results occurred among RT-qPCR-negative animals. Dogs and cats did not differ significantly in SARS-CoV-2 RT-qPCR positivity, ELISA reactivity, or Ct distributions. Because the ELISA had not been formally validated for the animal species studied, reactivity was not interpreted as confirmed SARS-CoV-2 seropositivity. The archived 2022–2023 dataset therefore provides a historical baseline for multispecies respiratory-virus surveillance in Kazakhstan and supports cautious integration of molecular and serological data.