Browse Articles

Long-tailed multi-label retinal disease classification using alternate group training and gradient-based re-weighting

Distinct tumor genomic signatures underlie canine macrophage polarization

Tumor-associated macrophages (TAMs) drive cancer progression and metastasis. However, the mechanisms by which tumor cells shape TAM phenotypes in canine cancers remain poorly understood. We investigated correlations between cancer cell gene expression and macrophage polarization to identify potential biomarkers and therapeutic targets. Tumor-conditioned media from 25 canine cancer cell lines were applied to monocyte-derived macrophages from three canine donors for 24 hours. Following washout, supernatants were analyzed for immunomodulatory cytokines and chemokines. Each cell line’s polarization capacity was ranked using modified z-scores, then correlated with RNA-sequencing data through Spearman’s correlation and differential expression analysis. Cancer cell lines showed marked heterogeneity in macrophage polarization capacity, largely independent of histologic type. MVB12A, a gene involved in exosome biogenesis, strongly correlated with vascular endothelial growth factor (VEGF) stimulation, suggesting exosome-mediated polarization mechanisms. Exosome fractionation experiments confirmed that purified exosomes induced significantly more macrophage VEGF secretion than other conditions, and high- MVB12A cell lines showed greater VEGF enrichment in exosomes. C-C motif chemokine ligand 3  ( CCL3) was strongly correlated with tumor necrosis factor-alpha (TNF-α) secretion exclusively in histiocytic sarcoma cells, and recombinant CCL3 induced dose-dependent TNF-α secretion from macrophages. High-polarizing cell lines exhibited upregulation of macrophage activation, epithelial-to-mesenchymal transition (EMT), and metabolic reprogramming genes, and downregulation of immune surveillance and cell adhesion genes. Gene set enrichment analysis confirmed pathways for immune suppression, EMT, and extracellular matrix remodeling. These findings identify exosome-associated VEGF stimulation as a previously uncharacterized mechanism in canine tumors and highlight CCL3 as a potential histiocytic sarcoma-specific driver of macrophage TNF-α secretion. Further validation in canine clinical cohorts will determine whether these pathways can serve as biomarkers or therapeutic targets in veterinary oncology.

Regional variation and prediction model of carbon emissions in the highway construction stage

Comparative emission and energy performance of biochar production versus in-situ burning of Gorean agrobyproducts

Uncontrolled open-field burning of agricultural residues remains a pervasive source of air pollutants and greenhouse gases in Korea, particularly for high-burning-ratio crops such as perilla, pepper, pear, grape, and apple residues. This study provides a comprehensive assessment of the environmental and energy implications of replacing open-field burning with biochar production from these major residue streams. First, mass- and energy-normalized emission factors for CO, CH₄, CO₂, NOₓ, N₂O, and particulate matter were quantified for both in-situ combustion and biochar combustion scenarios. Although biochar exhibited higher mass-based emission factors due to carbon and nitrogen enrichment during pyrolysis, the substantially lower mass required to deliver equivalent energy output resulted in markedly reduced annual emissions. System-level calculations showed that shifting from open-field burning to biochar conversion decreased CO, CH₄, and CO₂ emissions by 40.6–46.5%, NOₓ and N₂O emissions by 29.9–92.7%, and particulate emissions by 14.7–86.6%. Energy analysis demonstrated that biochar production is energetically favorable, with energy inputs of 2.65–3.50 MJ/kg and energy return on investment (EROI) values ranging from 8.57 to 10.06, exceeding the conventional viability threshold of 3.0. Apple pruning residues showed the highest biochar and net energy potentials, corresponding to 36.45 GWh/yr of electricity generation when co-fired in existing thermal power plants. Finally, greenhouse gas impacts were evaluated using 100-year global warming potential metrics. The biochar pathway reduced annual climate impacts by approximately 192,967 tCO₂eq/yr relative to baseline burning, translating to an estimated carbon-credit benefit of 4.44 billion KRW under 2020 market conditions. Overall, the findings indicate that biochar conversion provides a viable and impactful strategy for mitigating emissions and enhancing renewable energy recovery from agricultural residues in Korea.

Population-scale analysis of frequency-dependent calcium dynamics in retinal ganglion cells under electric field stimulation

Spatial distribution and determinants of unimproved sanitation facilities among households in Somalia: Using Somalia integrated household budget survey (SIHBS 2022)

Background Access to adequate sanitation remains a critical public health challenge in Somalia, where a large portion of the population relies on unimproved facilities due to persistent conflict, climate shocks, and political instability. This reliance contributes to a high burden of waterborne diseases. This study aimed to assess the spatial distribution of unimproved sanitation and identify its individual and community-level determinants using recent national data to inform targeted interventions. Methods This study is a secondary analysis of the 2022 Somalia Integrated Household Budget Survey (SIHBS), which included 7,212 households. The primary outcome was the use of unimproved sanitation facilities, categorized according to the WHO/UNICEF Joint Monitoring Programme (JMP) definitions. We employed a multilevel logistic regression model to identify individual and community-level determinants associated with unimproved sanitation. To analyze the spatial patterns of unimproved sanitation, we used Global Moran’s I for spatial autocorrelation and the Getis-Ord Gi* statistic for hotspot analysis. Results Overall, 36.87% of Somali households use unimproved sanitation facilities. There are significant disparities across residence types, with the highest prevalence among nomadic populations (83.28%), followed by rural (51.10%) and urban (23.88%) residents. The multilevel analysis revealed that households in permanent/formal housing (AOR: 3.42) and those with IDP status (AOR: 3.18) had significantly higher odds of using unimproved sanitation. At the community level, urban residence was paradoxically associated with higher odds of unimproved sanitation (AOR: 7.99) compared to rural areas, while nomadic populations had significantly lower odds (AOR: 0.04), likely reflecting a high prevalence of open defecation not captured as a “facility.” Spatial analysis identified significant hotspots of unimproved sanitation in the Hiraan (90.65%) and Bay (80.39%) regions, and cold spots in Banadir (5.37%) and Lower Shabelle (3.70%). Conclusion The findings highlight deep inequalities in sanitation access across Somalia, driven by geographic location, socioeconomic status, and population group. The high prevalence of unimproved sanitation, especially among nomadic, rural, and displaced populations, calls for urgent, geographically-targeted interventions. A multi-pronged approach is necessary, focusing on the specific needs of different communities and addressing the underlying structural and individual-level drivers of poor sanitation to advance public health and sustainable development goals in the region.

A pilot multiplex salivary transcriptomic analysis to understand the sex-specific effects of maternal opioid use in offspring

Gut microbiome alterations among Ghanaian children with asymptomatic malaria infections

The human gut microbiome, consisting of bacteria, archaea, fungi, and viruses, influences various physiological processes of the body. The gut microbiome composition is shaped by factors such as diet, geography, and antibiotic use. Malaria has been a global health challenge over the years, especially in low- and middle-income countries. This study investigated how asymptomatic malaria infection altered gut microbial communities in Ghanaian children, offering insights for novel malaria control strategies. Standard aseptic phlebotomy procedures were employed to collect venous blood samples for Plasmodium species detection. The gut microbial community was profiled by sequencing the 16S rRNA V4 region, and sequence data were processed using the DADA2 pipeline in R. Asymptomatic malaria infections were predominantly mixed with P. falciparum and P. malariae . Microbiome analysis revealed that Firmicutes and Bacteroidetes comprised nearly 70% of the total microbial population. Asymptomatic individuals showed a decrease in Firmicutes abundance from 52.5% to 44.0% and an increase in Bacteroidetes from 34.7% to 45.6%. There was also a slight increase in the abundance of Proteobacteria from 3.0% to 4.8%. At the genus level, Prevotella_9 was the most abundant and exhibited the highest variability in the infected groups. The Alloprevotella and Streptococcus genera increased in both infected groups, but Escherichia-Shigella was significantly elevated in only those with mixed infections. Faecalibacterium significantly declined in asymptomatic malaria-infected individuals compared to healthy controls, with variability further reduced in mixed infections. Beta-diversity analysis indicated a significant effect of malaria status on microbial composition (PERMANOVA, p < 0.05), explaining approximately 19.1% of the total variation captured by a 2D Principal Component Analysis (PCA) projection. These findings suggest a potential link between malaria infection and gut microbiota alterations and highlight microbial shifts associated with disease status.

Long-term kinetics of anti-RBD IgG antibodies 16 months after COVID-19 vaccination in Morocco: a longitudinal cohort study

Prevalence of cancer related fatigue and its associated factors among adult cancer patients in eastern Ethiopia: A cross-sectional study

Background Fatigue is a frequent and distressing symptom experienced by patients with cancer. It may result from the disease process and/or its aggressive treatment, which substantially impact the quality of life of cancer patients. Furthermore, there is a paucity of evidence in the Eastern part of Ethiopia. Hence this study aimed to assess the prevalence of cancer related fatigue and its associated factors among adult cancer patients, Eastern Ethiopia. Methods Hospital based cross-sectional study was conducted from 1 st May to 30 th August, 2023 among 422 systematically selected cancer patients. Data were collected using structured, interviewer administered questionnaire. The outcome variable was evaluated using the Brief Fatigue Inventory (BFI). Binary logistic regression analyses were performed to examine the association between the explanatory variables and the outcome variable. Adjusted Odds Ratio (AOR) with 95% Confidence Interval (CI) at a P value less 0.05 was used to declare statistically significant association. Results Out of 422, 382 individuals with various cancer types participated in the study with a response rate of 90.5%. The prevalence of cancer related fatigue was found to be 71.2% (95%CI: 65.7–75.5). Rural residence (AOR = 2.84, 95%CI: 1.25–6.43), female sex (AOR = 3.25, 95%CI: 1.49–7.0), private occupational status (AOR = 6.44, 95%CI: 2.42–17.12), never used coffee (AOR = 7.02, 95%CI: 2.37–20.75), inpatient admission (AOR = 4.68,95%CI: 2.21–9.88) and advanced (AOR = 6.21, 95%CI: 2.61–14.78) & unclassified cancer stages (AOR = 4.84, 95%CI: 1.42–16.57) were significantly associated with cancer related fatigue. Conclusions Nearly three out of four cancer patients in Eastern Ethiopia experienced cancer related fatigue. The findings highlight the need for a supportive care service including psychosocial counseling and targeted intervention for high risk groups such as female, rural residents, inpatients, private workers and advanced cancer patients. Further research is warranted to explore the protective role of coffee consumption.

Metaviromic analysis of Ixodes ticks in Northwestern Russia reveals high viral diversity and novel RNA virus lineages

Hierarchical coordinated scheduling algorithm for reactive power and voltage in cross-regional power grids based on multi-agent reinforcement learning

To address the challenges of strong dynamic coupling, action space dimension explosion, and voltage imbalance in reactive power and voltage scheduling of cross-regional power grids, this paper proposes a hierarchical coordinated scheduling method based on multi-agent reinforcement learning. The method first constructs a multi-agent reinforcement learning framework driven by probabilistic neural networks to perform distributed representation learning on the joint state vectors, achieving high-precision prediction of reactive power and voltage operating states for each node (prediction error MAE < 0.01 p.u.). Building upon the prediction results, a three-layer “prediction-decision-regulation” coordination mechanism is designed, integrating environmental state perception, action space optimization, and dynamic sensitivity analysis. This effectively addresses real-time decision-making challenges in high-dimensional action spaces, reducing average scheduling decision time by approximately 34.2%. Finally, sensitivity-driven feedback regulation achieves real-time balancing of reactive power and voltage at each node, guiding the power grid to converge stably to an optimal power flow state. Experimental results on the IEEE 33-node system demonstrate that the proposed method increases the voltage qualification rate to 98.7%, reduces system power loss by 30.5%, and decreases the maximum voltage magnitude deviation from 1.679 p.u. to 1.589 p.u., significantly outperforming traditional methods.

Combining machine learning and multi-omics analysis to explore the role of CPT1C in colorectal tumor cancer transformation

Stance modals in Chinese EFL learners’ monologue tasks: A corpus-based study

This study investigates how Chinese tertiary-level EFL learners used modals as linguistic devices for expressing stances in spoken English. Modals convey speakers’ commitment to propositions (epistemic stance) and intentions to affect reality (effective stance). While existing research has primarily focused on written registers and epistemic stances, this corpus-based study analyzes both stances in a national-level standardized speaking test, examining modals’ frequency, choices, and tonal variations during an opinion-giving task. Our findings reveal that learners commonly used epistemic and effective modals but leaned more toward effective modals than native speakers. Their speech exhibited assertiveness, marked by high use of will , should , and must while showing less reliance on would and have to . Moreover, can was often employed with conditional clauses/phrases to soften claims and used in rhetorical sentences to enhance assertiveness. These patterns could be attributable to the influence of the Chinese language, rhetorical norms, cultural influences, limited linguistic resources, and teaching support. Our findings add knowledge about modal stance in speaking tasks by revealing patterns of CEFLLs’ modal use during the period examined. These findings also offer implications for addressing learners’ reliance on certain modals.

Mechanisms underlying rhizosheath dynamics in Kengyilia hirsuta in response to alternating drought and rewatering

Abstract Under increasing frequency of extreme climate events, plant adaptation to alternating drought–rewatering stress is critical. Kengyilia hirsuta , a pioneer forage grass in alpine desert ecosystems, relies on rhizosheath formation for drought resistance. This study conducted indoor pot experiments with six water treatments: three drought–rewatering cycles (W1–W3, re‑watered to 10%, 25%, and 40% of field capacity, FC) and three sustained drought levels (W4–W6, maintained at 10%, 25%, and 40% FC). Root architecture, biomass allocation, arbuscular mycorrhizal fungi (AMF) colonization, and rhizosheath formation were examined over three successive 7‑day periods (T1–T3). Results revealed dynamic responses of rhizosheath accumulation to water regimes: maintained 25% FC (W5) significantly promoted rhizosheath biomass, maintained 40% FC (W6) enhanced early‑stage development, and re‑watering to 10% FC (W1) boosted later‑stage formation. AMF colonization increased progressively, with total colonization rising from 41.51% at T1 (day 7) to 61.40% at T3 (day 21). The W5 treatment consistently exhibited the highest vesicle, arbuscule, and hyphal colonization, along with increased soil spore density and hyphal density by T3. Root morphological traits—including tip number, volume, hair length, and hair density—also peaked under W5. Structural equation modelling identified AMF colonization (total effect: –0.90) and root hair traits (total effect: +0.80) as pivotal regulators of rhizosheath formation. This negative total effect of AMF colonization does not indicate overall inhibition, but rather reflects the feedback regulation intensity mediated by microbial competition and the carbon allocation trade-off within the plant-fungal symbiosis under resource-limited conditions. These factors interact through biomass allocation, root architecture, and soil microenvironment, forming a multidimensional adaptive network. These findings elucidate the ecophysiological mechanisms of plant–AMF collaboration in rhizosheath formation under water fluctuation, supporting the selection of stress‑tolerant grasses for restoring desertified grasslands.

Analysis of catabolic products of L-arginine; L-ornithine and L-citrulline and the residual L-arginine using the HPLC and LC-MS

L-arginine, a semi-essential amino acid, is metabolised in the cell to generate nitric oxide (NO) and L-citrulline via the enzyme nitric oxide synthase (NOS) or urea and L-ornithine via arginase activity. L-citrulline and L-ornithine are the products of L-arginine degradation. Mouse liver epithelial (BNL CL2) and mouse embryonic fibroblast (3T3 L1) insulin-sensitive cell lines were used as model systems and cultured with 0, 400 or 800 µM L-Arg. This study focuses on the analysis of the residual concentrations of amino acids (L-Arg, L-Cit and L-Orn) in cell culture medium samples using high performance liquid chromatography that involves precolumn derivatization with o-phthaldialdehyde. In BNL CL2 cells, most of the culture supernatant has increased amount of L-Arg in comparison to the control complete DMEM addition. L-ornithine levels showed an overall increase over time, with higher concentrations observed at 72 h compared with 24 h across all samples. In 3T3 L1 cells, residual L-Arg concentration decreased in most of the cell supernatant in comparison to the control at 72 h. Noticeably, L-Arg at 0 µM and the control complete DMEM had highest amount of L-Orn among all samples. Interestingly, L-Cit was very much high in culture medium of both untreated BNL CL2 (85.96 µM) and 3T3 L1 (37.49 µM) cells at T = 0 compared to the control. Collectively, the results show that excess L-Arg is sensed by the cell which then regulates the residual amount of amino acids concentration. The spectroscopy technique used here is highly sensitive, specific and accurate, can be readily automated and serves as a valuable tool for investigating the modulation of the arginine-nitric oxide pathway.

Assessing bone radiodensity and thickness in cochlear implant patients through manual photon-counting CT image segmentation using ITK-SNAP

Abstract Bone radiodensity and thickness between the cochlear implant’s middle electrodes and the facial nerve were compared across four groups—Facial nerve stimulation (FNS) patients with and without far advanced otosclerosis (FAO), and controls with and without FAO—using manual segmentation of photon counting-computed tomography (PC-CT) to assess FNS risk. This case-control study compared FAO patients with FNS ( n  = 3) to non-FAO with FNS ( n  = 2) and two control groups without FNS of post-lingually deafened patients: FAO ( n  = 2) and non-FAO ( n  = 2). Clinical data from medical record included surgical details, complications, PC-CT findings (bone thickness between the middle electrodes and the facial nerve), hearing outcomes, and implant fitting at two years. Manual segmentation of PC-CT was performed using ITK-SNAP software to measure the bone radiodensity between the mid-array electrode and the adjacent facial nerve. No significant differences were found between FNS and non-FNS patients in demographics, surgical outcomes, complications, audiometric data, or implant programming at 2 years post-surgery. A significant difference in bone radiodensity was observed in patients with FAO (1483.57 ± 122.37 HU (Hounsfield Units)) compared to those without FAO (2403.51 ± 128.24 HU; p  < 0.001). In the non-FAO with FNS group, one patient exhibited an exceptionally short distance (0.13 mm) between the first portion of the facial nerve and the middle electrodes. When comparing FNS and non-FNS patients, no significant differences were observed in bone radiodensity ( p  = 0.85) or in bone thickness ( p  = 0.75). ITK-SNAP enables manual PC-CT segmentation to assess bone radiodensity and thickness between cochlear implant electrodes and the facial nerve. In otosclerosis, lower radiodensity alone doesn’t explain FNS, but reduced bone thickness may contribute.

PlantaNet and PlantaNetLite: Efficient and explainable multi-crop plant disease classification via transformer benchmarking and custom lightweight CNNs

Plant disease diagnosis based on visual symptoms is crucial for preventing yield loss; however, deployment in practical settings remains challenging due to inter-class similarity, background noise, and limited computational resources. This study presents a plant disease classification framework evaluated on a curated multi-crop dataset aggregated from multiple publicly available repositories, comprising 51 disease and healthy classes. The dataset includes approximately 45,000 original images that were expanded through controlled augmentation during training to improve generalization. We benchmark eight ImageNet-pretrained tiny vision transformer architectures trained for up to 50 epochs. Among these, CAFormer-s18 achieved strong validation performance but with increased computational overhead. To enable efficient and computationally lightweight solutions, we design two fully customized convolutional neural networks: PlantaNetLite (1.28M parameters) and PlantaNet (2.58M parameters). After hyperparameter optimization and full 100-epoch training, PlantaNet achieved 99.37% validation accuracy and 99.66% test accuracy with a compact model size (9.85 MB) and moderate computational cost, while PlantaNetLite achieved a best validation accuracy of 99.22% under further parameter reduction. Qualitative Grad-CAM and Grad-CAM++ analyses provide insight into the regions influencing model predictions. Overall, the proposed models demonstrate competitive accuracy while maintaining computational efficiency, highlighting their potential suitability for resource-constrained deployment scenarios.

Real-life multicenter experience of long-term treatment with venetoclax plus azacitidine for acute myeloid leukemia in China

Development and validation of the MIPPE: A novel dyadic assessment tool for early parent-child interactions in clinical practice

Early parent-child interactions are crucial for child development. Existing assessment scales have several limitations: intensive training often exceeding 40 hours, administration time up to two hours, and unbalanced distribution of items to the detriment of the dyadic dimension of interactions. Our study aims to develop and validate the MIPPE (Measure of Early Parent-Child Interactions), a scale adapted for daily clinical practice that assesses the quality of early interactions while integrating the dyadic dimension. The MIPPE scale was developed within the PERL program in Eastern France. After several revisions, the final version includes 9 items scored from 0 to 3. Validation is based on the analysis of 228 parent-child interaction videos from 123 dyads at 4 and 24 months, divided between intervention (N = 62) and control groups (N = 61). The MIPPE demonstrates acceptable internal consistency (Cronbach’s α = 0.92, McDonald’s ω = 0.93). Exploratory factor analysis reveals a unidimensional structure explaining 66% of the variance, confirmed by confirmatory factor analysis (CFI = 1.000, TLI = 1.002, RMSEA = 0.000). Inter-item correlations range from 0.34 to 0.67, indicating satisfactory cohesion. Clinical thresholds have been established: 15 and 20 (optimal sensitivity 87%). The MIPPE constitutes an accessible, rapid, and psychometrically validated assessment tool for early childhood professionals. It facilitates early screening of interactional difficulties and referral to interventions, thus contributing to the prevention of developmental disorders and parental support.