Advancing Open Science

Recommender systems are widely used in e-commerce to alleviate information overload, yet the nonlinear decision mechanisms of neural recommenders make it difficult to verify whether an identified feature is genuinely responsible for a recommendation. To address this problem, this study proposes a counterfactual-empowered causal analysis and trustworthiness verification framework that operates on interpretable semantic attributes extracted from review text. DeepSeek is used as a constrained semantic annotator to extract aspect–opinion–sentiment triplets from Amazon Cell Phones and Accessories reviews, thereby constructing an explicit semantic intervention space. A symmetric multilayer perceptron (MLP) is trained as the recommendation predictor and is kept fixed during verification; the verifier assumes gradient access with respect to the semantic input representation but does not update the predictor parameters. Three mechanisms are introduced to improve counterfactual verification: (1) AdamW optimization with cosine-annealing scheduling for stable continuous perturbation search; (2) a gradient-guided greedy strategy that maps continuous perturbations to discrete human-understandable semantic attributes; and (3) Elastic Net regularization that jointly constrains explanation sparsity and perturbation magnitude. Experiments on the processed Amazon Cell Phones and Accessories dataset show that the proposed components improve Precision, Probability of Necessity (PN), Probability of Sufficiency (PS), and Feature Necessity Score (FNS) relative to the evaluated optimization and ablation variants. These results indicate that semantic counterfactual intervention provides a practical post-hoc mechanism for assessing whether recommendation rationales are causally effective rather than merely correlated with model outputs.

Background: Multidrug-resistant Klebsiella pneumoniae (MDR-KP) is a global health threat with limited treatment options. Comparative in vitro data for imipenem-relebactam (I/R) and ceftolozane-tazobactam (C/T) across diverse MDR-KP clinical populations remain limited. Methods: Susceptibility to I/R and C/T was determined by Etest. Whole-genome sequencing was used to identify sequence types, resistance genes, and virulence genes, with hypervirulent K. pneumoniae (hvKP) defined by genomic criteria. Results: A total of 312 nonduplicate MDR-KP clinical isolates, including 70.8% carbapenem-resistant K. pneumoniae (CR-KP), 56.7% difficult-to-treat resistant K. pneumoniae (DTR-KP), and 28.2% carbapenem-resistant hvKP (CR-hvKP), were retrospectively analyzed. I/R resistance was low across all subgroups (overall 3.8%, CR-KP 5.0%, DTR-KP 6.2%, CR-hvKP 2.3%) and across specimen types (0–9.6%; blood 3.8%). In contrast, C/T resistance was higher (overall 57.7%, CR-KP 79.2%, DTR-KP 91.0%, CR-hvKP 84.1%), with the highest rate in respiratory isolates (63.0%), followed by blood (57.7%) and urine (50.0%). Among extended-spectrum β-lactamase (ESBL)-producing isolates, I/R and C/T resistance rates were 2.1% and 12.4%, respectively. Although C/T resistance was higher in hvKP than in classic K. pneumoniae in unadjusted analysis, hvKP was not independently associated with increased C/T resistance after adjustment. Instead, ST11, ST15, ST307, bla_KPC-2, and bla_NDM-1 were independently associated with C/T resistance. For I/R, bla_NDM-1 was associated with resistance. Conclusions: I/R was active across MDR-KP subgroups including CR-hvKP and DTR-KP. C/T worked mainly against ESBL-producers. These findings highlight the importance of resistance genotyping when interpreting susceptibility to these agents.

Background/Objectives: Establishing drowning as the cause of death remains one of the most challenging tasks in forensic pathology, particularly when classical autopsy findings are absent, nonspecific, or compromised by decomposition. Molecular biomarkers may provide objective adjunctive evidence in such cases. This study investigated the diagnostic utility of Aquaporin-5 (AQP5) and Surfactant Protein A2 (SP-A2) measured in renal tissue as potential biomarkers of freshwater drowning and evaluated the influence of different freshwater environments on their expression. Methods: A controlled experimental study was performed using 29 adult Wistar rats allocated to freshwater drowning groups (Boračko Lake, n = 11; Jablaničko Lake, n = 11) and non-drowning control groups subjected to mechanical asphyxia followed by one hour of postmortem immersion (n = 7). Kidney tissues were collected within a maximum of 5 min after death and analyzed for AQP5 and SP-A2 concentrations using enzyme-linked immunosorbent assay (ELISA). Group differences were evaluated using Kruskal–Wallis and Dunn’s post hoc tests, while effect sizes were estimated using Cliff’s delta. Receiver operating characteristic (ROC) curve analysis was performed to assess the diagnostic performance of each biomarker in distinguishing drowning from non-drowning deaths. Results: AQP5 concentrations differed significantly among groups (χ² = 16.05, p = 0.00033), with large effect sizes observed in all pairwise comparisons. However, diagnostic performance was poor, as ROC analysis demonstrated no significant discrimination between drowning and non-drowning cases (AUC = 0.534, 95% CI: 0.314–0.754; p = 0.762). In contrast, SP-A2 showed significant intergroup differences (χ² = 6.69, p = 0.035) and moderate diagnostic accuracy for drowning identification (AUC = 0.752, 95% CI: 0.567–0.937; p = 0.008), with a sensitivity of 77.2% and specificity of 81.8% at the optimal threshold. Notably, AQP5 expression appeared suggestive of environmental modulation, whereas SP-A2 demonstrated greater relevance to the drowning process itself. Conclusions: Renal SP-A2 showed moderate diagnostic utility as a supportive biomarker for drowning, whereas AQP5 lacked sufficient discriminatory power for cause-of-death determination. The marked variability of AQP5 across freshwater environments suggests that it reflects environmental modulation rather than drowning-specific biological responses. Although neither marker is suitable as a standalone diagnostic tool, renal SP-A2 may represent a useful adjunct in forensic drowning investigations, particularly when pulmonary tissue is unavailable or unsuitable for analysis. These findings should be considered preliminary and applicable only to freshwater drowning conditions, as saltwater drowning was not investigated in the present study. Further validation in human forensic casework and larger experimental cohorts is warranted.

Rosemary (Rosmarinus officinalis L., syn. Salvia rosmarinus Spenn., Lamiaceae) is a Mediterranean perennial herb with high tolerance to environmental stress, but its long-term response to combined drought and salinity and the potential of biostimulants to modulate acclimation remain insufficiently understood. This study evaluated the effects of chronic drought and salinity, applied individually or in combination, on rosemary growth, water status, ion homeostasis, physiological stability, and essential oil composition, and assessed the mitigating effects of foliar nano-chitosan and seaweed extract. The experiment was conducted under arid Mediterranean conditions over two consecutive years (2024–2025) with four harvests, using a split–split–plot randomized complete block design with three irrigation levels, three salinity levels, and three biostimulant treatments. Drought, salinity, and their interaction significantly affected all evaluated traits. Severe combined stress reduced dry weight by 33.3%, while overall physiological performance remained relatively stable. Under severe combined stress, B2 treatment increased relative water content and the K⁺/Na⁺ ratio by 14.9% and 66.7%, respectively, and reduced Na⁺ accumulation by 17.9%. B2 showed a consistently higher Biostimulant Efficiency Index than B1, indicating a greater response to the higher-concentration combined formulation. Moderate drought with B2 increased essential oil contents of camphor, a-pinene, and 1,8-cineole. Overall, rosemary maintained relatively stable physiological performance despite significant stress-induced changes in individual traits, indicating a degree of physiological resilience to chronic combined stress while the nano-chitosan–seaweed formulation was associated with improved water relations and ionic balance under severe combined stress, although these changes did not translate into biomass recovery.

Scleranthus perennis L. is an Eurasian species from Caryophyllaceae family, Sclerantheae tribe. The work aimed at investigating the profiling of secondary metabolites in S. perennis aerial parts by means of ultra-high-performance liquid chromatography–high-resolution mass spectrometry (UHPLC–HRMS). The profile was delineated by phenolic compounds with a γ-pyrone core, including 49 C-glycosyl flavones O-glycosylated on the sugar of the C-glycosylation and their acetyl, galloyl, and methoxycinnamoyl esters alongside di-C-flavone glycosides. For the first time, a variety of maltol-glycosides was annotated together with phenolic acids and their glycosides. The total phenolic and flavonoid contents were 27.7 ± 0.4 mg gallic acid equivalents/g dry extract (de) and 31.6 ± 0.1 mg rutin equivalents/g de. Moderate ABTS radical scavenging activity (IC₅₀: 1.21 mg/mL) and reducing power in CUPRAC (EC₅₀: 1.40 mg/mL) and FRAP (EC₅₀: 0.61mg/mL) assays were observed. Pronounced inhibitory activity towards α-glucoside (IC₅₀: 1.03 mg/mL) and acetylcholinesterase (IC₅₀: 1.45 mg/mL) was established. Our results are a prerequisite for further phytopharmacological and chemophenetic studies.

Hyperspectral imaging (HSI) combines spatial and spectral information and has been widely investigated for nondestructive sensing in agriculture and food. This review examines HSI applications in crop growth monitoring, abiotic and biotic stress detection, soil assessment, and postharvest quality and safety inspection. Particular attention is given to how acquisition configuration, spectral preprocessing, feature extraction, and machine- or deep-learning models affect performance across laboratory, field, and production-line settings. Recent studies show that red-edge and shortwave-infrared features are useful for estimating chlorophyll, nitrogen, and water status, while learning-based models have been applied to disease detection, contaminant identification, and quality grading. However, transfer across years, sites, and instruments remains difficult, and practical deployment is further limited by calibration inconsistency, hardware cost, and operational complexity. We therefore discuss instrument-aware calibration and preprocessing, model generalization, lower-cost sensing, and deployment strategies that may improve reproducibility and field applicability.

Background/Objectives: Ethanol-induced gastric injury is a widely used experimental model for evaluating gastroprotective interventions because it reproduces key features of gastric mucosal damage, including oxidative stress, lipid peroxidation, inflammation, and epithelial disruption. Previous phytochemical characterization of Montanoa bipinnatifida leaves revealed a rich profile of bioactive polyphenols with antioxidant potential. Therefore, this study investigated the gastroprotective effects of a hydroethanolic leaf extract of M. bipinnatifida in female CD-1 mice subjected to ethanol-induced gastric injury. Methods: Animals received the extract orally for seven consecutive days before administration of 50% ethanol. Gastric tissues were subsequently evaluated by macroscopic, histopathological, biochemical, and phytochemical analyses. Results: Ethanol exposure induced severe gastric damage characterized by hemorrhagic lesions, epithelial degeneration, mucosal edema, and inflammatory cell in-filtration. In contrast, pretreatment with the extract significantly reduced gastric injury, preserving tissue architecture, and attenuating histopathological alterations. Biochemical and molecular analyses revealed a marked decrease in malondialdehyde levels and IL-1β expression following extract pretreatment, whereas TNF-α expression was not significantly reduced; SOD2 and catalase protein expression showed no significant treatment-associated changes. HPLC-DAD analysis confirmed the presence of catechin and procyanidin B2 in the extract. Conclusions: Overall, these findings indicate that M. bipinnatifida leaf extract attenuates acute ethanol-induced gastric injury and that this gastroprotective response is associated with reduced lipid peroxidation and selective attenuation of IL-1β rather than generalized suppression of the inflammatory markers examined. These results support further investigation of M. bipinnatifida as a potential source of gastroprotective bioactive compounds.

To tackle frequent voltage violations in distribution networks with high photovoltaic penetration ratios, voltage regulation methods that dispatch the active and reactive powers of PV panels are widely studied. Because of the costly global communication networks, in current distributed voltage regulation methods based on multi-agent reinforcement learning, each agent determines control actions according to its individual observation, which implies a partially observable Markov decision process. However, the determined control actions are far from optimal, resulting in less active power generated by PV panels and voltage violations. Some recent voltage regulation methods, where each agent determines control actions according to its historical data of the most recent time steps, can improve the optimality, but the improvement is still not enough. This paper proposes a spatiotemporal multi-agent reinforcement learning voltage regulation (ST-MARL-VR) method where each agent determines control actions according to its historical observations and those of the neighboring agents. Historical observations of neighboring agents are transferred through a point-to-point communication network, which can largely reduce the costs for communication network establishment and improve the communication robustness compared to global communication networks. Furthermore, in practice, our proposed method can be deployed on the terminal units in distribution networks. Subsequently, we validate the proposed ST-MARL-VR method under two test scenarios, and the quantitative performance improvements are above 20% compared with the two comparative methods. Comparative analyses against state-of-the-art benchmarks demonstrate that the developed ST-MARL-VR algorithm substantially enhances the optimality of the determined control actions.