Browse Articles
Discover research articles across all indexed journals
Food safety practice and associated factors among food handlers working in food and drinking establishments in Debre Birhan City, North Eastern Ethiopia: A convergent parallel mixed-method study
Background Food-borne illnesses pose a major global public health threat, affecting up to 30% of people annually in higher-income countries and contributing to hundreds of thousands of deaths each year with the heaviest burden falling on low and middle income countries. In Ethiopia, prior research on food handlers’ safety practices has been limited to quantitative methods offering little insight into cultural, behavioral and institutional barriers. To the best of our knowledge, limited mixed-methods evidence exists in this context. Therefore, this mixed-methods study addresses the gap through providing comprehensive evidence from North Eastern Ethiopia on factors influencing food safety practices. Objective This study aims to assess food safety practices and associated factors among food handlers working in food and drinking establishments in Debre Birhan city, North Eastern Ethiopia. Methods A convergent-parallel mixed-methods study with a cross-sectional quantitative strand was conducted from January 10 to February 28, 2025, among 415 randomly selected food handlers in Debre Birhan city, North Eastern Ethiopia. Quantitative data were collected using a pre-tested structured questionnaire and observation method and analyzed using STATA version 14. Multivariable binary logistic regression with 95% confidence intervals and a p-value ≤ 0.05 was used to identify factors associated with food safety practices. Qualitative data were analyzed using ATLAS.ti version 8 software. Results The magnitude of poor food safety practices among food handlers was 72%. Educational status (AOR = 3.2; 1.206–5.337), working ≤ 2 years (AOR = 2.7; 1.026–4.009), lack of food safety training (AOR = 4.3; 1.997–7.820) and poor knowledge (AOR = 3.9; 1.853–6.983) were statistically associated with poor food safety practice among food handlers. Lack of food safety training, poor risk perception, weak enforcement, lack of standardized guidelines, inadequate facilities and insufficient equipment were identified as barriers for good food safety practice. Conclusions Food safety practice among food handlers in the current study was low. Quantitative findings showed that educational status, working ≤ 2 years, lack of food safety training and poor knowledge were factors statistically associated with poor food safety practice. Qualitative findings provided limited access to training, poor risk perception, weak enforcement, lack of standardized guidelines, inadequate equipment and facilities jointly undermine food safety compliance. Addressing these factors particularly through enhanced training and knowledge improvement as well as regular and supportive supervisory, monitoring and evaluation mechanisms could substantially improve adherence to safe food handling and reduce the risk of foodborne diseases.
A mechanism for adaptive genome regulation in cancer
Automated, physics-guided AI framework for asymmetry-aware ferroelectric compact models
Mukara: A deep learning alternative to the four-step travel demand model with a case study on interurban highway traffic prediction in the UK
Accurate traffic volume prediction is essential for managing congestion, improving road safety, mitigating environmental impacts, and supporting long-term transportation planning. The traditional four-step travel demand model (FSM) is a well-established framework, but it relies on static survey data, substantial calibration effort, and simplified behavioural assumptions that may not adequately capture complex travel patterns. In contrast, data-driven models are capable of learning nonlinear relationships from large datasets, yet they are often designed for short-term forecasting and typically do not target the long-term, segment-level volume estimation tasks required for strategic planning. This study proposes Mukara, a deep learning framework that directly approximates the mapping from external socioeconomic and network features to observed traffic volumes on highway trunk road segments. The model is trained on eight years of data from England and Wales and incorporates population, employment, land use, road network characteristics, and points of interest as inputs. Mukara achieves a mean GEH of 50.74, a mean absolute error of 8,989 vehicles per day, and an R 2 of 0.583 under random cross-validation, outperforming baseline models and existing studies under comparable settings. Under a more stringent region-based spatial cross-validation scheme, performance remains robust, demonstrating strong spatial transferability. Ablation experiments further demonstrate the robustness of the proposed architecture and reveal the relative importance of different input feature groups for prediction.
SLAMF6 as a drug-targetable suppressor of T cell immunity against cancer
Retinal arteriovenous crossings as spatiotemporal stress fields: a quantitative Doppler OCT flowmetry study
Lack of tetrodotoxin analogues and individual metabolomic profiling of the cryptic frog Colostethus imbricolus
Poison frogs (Dendrobatoidea) are characterized by the great diversity of alkaloids discovered in their skin. However, most of these alkaloids have been found in brightly colored species and there is a wide lack of knowledge of alkaloid profiles in the less colorful species. Previous finding of paralytic tetrodotoxins (TTXs) in only two cryptically colored species from the genus Colostethus, establishes the unique occurrence of hydrophilic alkaloids in the superfamily Dendrobatoidea. Unpublished results using extracts from Colostethus imbricolus , demonstrated that this species contains paralysis-producing substances, after intraperitoneal injection of mice. To analyze their skin metabolites and to determine if they correspond to TTX, or TTX analogues, we have employed a TTX-targeted separation in normal phase gradient, and an untargeted profiling in reversed-phase gradient. After performing both analyses, neither TTX nor TTX-analogues were detected in C. imbricolus . In contrast, other metabolites were separated, allowing the extraction of 76 adducts common to both analyses, being 33 of them tentatively annotated as amphibian alkaloids, eight as amphibian metabolites different from alkaloids and 25 that matched with natural products from the DNP. A total of 10 common molecular formulas remained non-annotated. The absence of MS/MS spectra for these adducts requires their structures to be confirmed in future analyses, following the completion of targeted MS/MS acquisition. After analyzing the inter-individual variation of six specimens, it was demonstrated that the skin metabolome differs between males and females of C. imbricolus . Our results lead us to conclude that TTX is not the only paralyzing compound in dendrobatid frogs and that more work should be undergone to identify this phenomenon. A notable additional outcome of this study is the first successful separation of TTX on an SB-CN column using a normal-phase gradient, suggesting a potential useful approach for TTX-targeted separation.
Language models transmit behavioural traits through hidden signals in data
Abstract Large language models (LLMs) are increasingly used to generate data to train improved models 1–3 , but it remains unclear what properties are transmitted in this model distillation 4,5 . Here we show that distillation can lead to subliminal learning—the transmission of behavioural traits through semantically unrelated data. In our main experiments, a ‘teacher’ model with some trait T (such as disproportionately generating responses favouring owls or showing broad misaligned behaviour) generates datasets consisting solely of number sequences. Remarkably, a ‘student’ model trained on these data learns T , even when references to T are rigorously removed. More realistically, we observe the same effect when the teacher generates math reasoning traces or code. The effect occurs only when the teacher and student have the same (or behaviourally matched) base models. To help explain this, we prove a theoretical result showing that subliminal learning arises in neural networks under broad conditions and demonstrate it in a simple multilayer perceptron (MLP) classifier. As artificial intelligence systems are increasingly trained on the outputs of one another, they may inherit properties not visible in the data. Safety evaluations may therefore need to examine not just behaviour, but the origins of models and training data and the processes used to create them.
SLC30A10 downregulation is associated with cGAS-STING pathway activation in colorectal tubular adenoma
Abstract Based on data-independent acquisition (DIA) proteomics technology, to analyze the proteomic characteristics of colorectal tubular adenoma and explore the expression changes of SLC30A10 and their potential association with the cGAS-STING pathway. A self-controlled design was adopted, collecting colorectal tubular adenoma (TA) and paired normal mucosa (NM) from 15 patients with TA. Differentially expressed proteins were screened by DIA proteomics, followed by GO and KEGG enrichment analyses. Immunohistochemistry was performed to detect the expression of SLC30A10, cGAS, STING, p-IRF3, ISG15, and β-catenin; immunofluorescence double staining was used to observe the co-localization of p-IRF3 and β-catenin; inductively coupled plasma mass spectrometry (ICP-MS) was employed to determine tissue manganese content. DIA analysis showed that SLC30A10 protein expression was significantly downregulated in TA tissues. Functional enrichment analysis indicated abnormalities in signal transduction, metabolic reprogramming, and nitrogen metabolism in TA tissues. IHC results demonstrated that, compared with NM, TA tissues exhibited reduced expression of SLC30A10, while the expression of cGAS, STING, p-IRF3, ISG15, and β-catenin was upregulated. Manganese content in TA tissues was also significantly increased. Immunofluorescence revealed enhanced nuclear signals of p-IRF3 in TA cells, with co-localization of p-IRF3 and β-catenin observed in the nucleus. Downregulation of SLC30A10 in colorectal tubular adenoma is associated with manganese accumulation and alterations in the cGAS-STING pathway, suggesting its potential role in the development and progression of adenoma, a finding with promising research implications.
Statistically valid explainable black-box machine learning: applications in sex classification across species using brain imaging
Sex classification using neuroimaging data has the potential to revolutionize personalized diagnostics by revealing subtle structural brain differences that underlie sex-specific disease risks. Despite the promise of machine learning, traditional methods often fall short in providing both high classification accuracy and interpretable, statistically validated feature importance scores for high-dimensional imaging data. This gap is particularly evident when conventional techniques such as random forests, LIME, and SHAP are applied, as they struggle with complex feature interactions and managing noise in large datasets. We address this challenge by developing an integrated framework that combines Oblique Random Forests (ORFs) with a novel, permutation-based feature importance testing algorithm. ORFs extend traditional random forests by employing oblique decision boundaries through linear combinations of features, thereby capturing intricate interactions inherent in neuroimaging data. Our feature importance testing method, NEOFIT, rigorously quantifies the significance of each feature by generating null distributions and corrected p -values. We first validate our approach using simulated datasets, establishing its robustness and scalability under controlled conditions. We then apply our method to classify sex from both voxel-wise structural MRI and cortical thickness data in humans and macaques, facilitating direct cross-species comparisons. ORFs achieves AUC > 0.80 on human data, and >0.70 on macaque data, while NEOFIT identifies statistically significant features aligned with sex-dimorphic neuroanatomy. Our results demonstrate that the proposed framework not only enhances classification performance but also provides clear, interpretable insights into the neuroanatomical features that distinguish sexes. These methodological advancements pave the way for improved diagnostic tools and contribute to a deeper understanding of the evolutionary basis of sex differences in brain structure.
Microbial diversity and production of milk spirit using traditional Buryat fermentation and distillation technologies
Abstract Distilled fermented milk beverages are rare in food technology, despite the global prevalence of plant-based spirits. Currently, the production of distilled strong alcoholic beverages from fermented milk using traditional technologies is known only among Mongolic-speaking peoples and their Siberian neighbors. This study provides the first interdisciplinary analysis of darasun , a traditional Buryat spirit made from fermented milk beverage khurenge in Southern Siberia. Archival research shows deep historical roots of darasun dating back to the 13th–14th centuries in the period of the Mongol Empire. Low content of fermentable carbohydrates in milk limits the maximum ethanol concentration in the fermented product, which reaches 1.5–2.1%. To increase the beverage’s strength, a traditional multi-stage distillation is used, allowing sequential increases in ethanol concentration to 11%, 40%, and 75%. ITS and 16S rRNA gene sequence analysis identified diverse microbial community in khurenge dominated by representatives of Lactobacillus genus among lactic acid bacteria and Trichosporon or Monosporozyma genera among yeasts. Comparison of home-produced and industrially produced khurenge revealed that pasteurization significantly affects microbial composition with non-pasteurized khurenge showing greater microbial diversity and lower ethanol content than pasteurized. Functional analysis showed that pathway composition differed significantly between samples suggesting that final ethanol concentration in fermented milk reflects the balance between competing fermentation and respiration pathways. This study demonstrates that Buryat milk fermentation and distillation represents a unique biotechnological system that exemplifies decentralized, low-input, circular bioeconomy principles adapted to pastoral settings, providing a contrast to industrial whey-to-ethanol processes and contributing to the preservation of endangered cultural heritage.
Pooled incidence and predictors of infant mortality in low- and middle-income countries using gamma shared frailty model: Insights for achieving the Sustainable Development Goals
Background Low- and middle-income countries (LMICs) account for a large share of global infant deaths, but there is a lack of evidence on the pooled estimate of infant mortality and its predictors in LMICs. Therefore, this study aimed to assess the pooled incidence of infant mortality and its associated factors in LMICs. Methods We used clustered data extracted from the recent Demographic and Health Surveys (DHS 2018-DHS 2024) of all LMICs. A total of 1,404,826weighted numbers of recent live births were included in the study. A lognormal shared gamma frailty model was employed. We used the Akaike information criterion (AIC), Bayesian information criterion (BIC), and log-likelihood values for model comparison. An adjusted time ratio ( ϕ ) with a 95% confidence interval (CI) in the final model was used to select variables that had a significant association with time to infant death. The data were analyzed via R software version 4.3.1. Results A total of 1,404,826 live births were included in the final analysis. By the end of the follow-up period, 72,569 infants (5.17%, 95% CI: 5.13–5.21) had died before their first birthday. The pooled estimate of the IMR in LMICs was 39 per 1000 live births (95% CI: 32.68–44.95). Maternal education, family size ≥ 5, being a multiparous mother, being delivered at health facilities, being a female infant, immediate initiation of breast feeding, living in Europe & Central Asia, and living in West & East Asia were significantly associated with a lower risk of infant death. Conversely, maternal age 25–34, maternal age 35–49, unimproved toilet facilities, poor and middle wealth indices, maternal age at birth ≤19, birth interval of <18 and 18–23 months, multiple births, 2 nd birth order, small birth size, low and medium Human Development Index (HDI), low and medium literacy rate, low-income and lower-middle income countries, rural residence, living in West Africa, South & Central Africa, and South Asia were significantly associated with a higher risk of infant mortality. Conclusion The infant mortality rate (IMR) in LMICs remains high compared with that in WHO targets and shows significant regional variation. West Africa and South Asia had the highest pooled estimate of infant deaths. Variables such as maternal age, education, wealth index, age at first birth, parity, family size, child sex, birth interval, multiple pregnancy, birth order number, perceived child size at birth, place of delivery, residence, country’s literacy rate, income group, and HDI value were identified as significant predictors of time to infant death. Therefore, public health interventions that enhance health facility delivery, optimal birth spacing, maternal education, and immediate breastfeeding are crucial to reduce the incidence of infant mortality in LMICs.
Bio-inspired chemometric methods for simultaneous UV spectrophotometric determination of molnupiravir, nirmatrelvir, and favipiravir in pharmaceutical formulations and environmental samples
Academic stress and its psychosocial and behavioral determinants in medical students: Findings from a cross-sectional study
Background Academic stress is a widespread challenge in medical education, with psychological, behavioral, and contextual factors contributing to it. This study estimated the prevalence of academic stress among Bangladeshi medical students and identified key psychosocial and behavioral predictors to guide targeted interventions. Methods A multicenter cross-sectional study (October–December 2022) used a stratified random sample of 1,072 undergraduate students from eight public medical colleges representing all administrative divisions of Bangladesh. Validated instruments measured academic stress (Academic Stress Scale, ASS-40), depressive symptoms (PHQ-9), anxiety (GAD-7), insomnia (ISI), internet addiction (IAT), self-esteem (RSES), and coping styles (SCSI). Analyses included descriptive statistics, chi-square and Mann–Whitney U tests, multivariable logistic regression to identify independent predictors, and structural equation modeling (SEM) and network analysis to explore direct and indirect pathways. Result Academic stress was reported by 47.5% of participants. In adjusted logistic regression models, moderate anxiety was associated with increased odds of academic stress (AOR = 3.95; 95% CI 1.98–7.90), and severe depression showed a markedly elevated association (AOR = 21.54; 95% CI 7.21–64.38). Behavioral factors were also influential: moderate-to-severe problematic internet use was strongly associated with academic stress (AOR = 17.78; 95% CI 9.66–32.72). Additional independent predictors included advanced academic year, higher monthly expenditure, and urban residence. Active problem-focused coping conferred modest protection against academic stress (AOR = 0.89; 95% CI 0.83–0.95). Structural equation modeling supported a model in which psychological distress exerted both direct effects on academic stress and indirect effects mediated by sleep disturbance and internet addiction, while network analysis identified depressive symptoms, insomnia, and internet addiction as central nodes within the stress network. Conclusions Nearly half of the sampled medical students experienced significant perceived academic stress. Interventions that integrate mental health services, sleep-hygiene promotion, responsible digital-use policies, and training in adaptive, problem-focused coping are recommended.
Investigation of the incidence of early and delayed postoperative nausea and vomiting (PONV) in different surgical specialties
“Because I said so.” – Collection and evaluation of parenting phrases in German-speaking samples
Verbal communication is a key aspect of parenting behaviour. However, phrases frequently used in parenting situations (e.g., “Stop crying”, “I love you”) have yet to receive focused attention in the research literature. This set of two online questionnaire studies investigated the use of parenting phrases within a German-speaking sample. The first study compiled an inventory of common parenting phrases, contextualized them within parenting style research on the dimensions of warmth and control, and examined their associations with self-esteem. A total of 309 phrases were collected, of which the 84 most frequently mentioned were further analysed in Study 2, which examined their ratings within a communication theory perspective and investigated socio-demographic influences on participants’ evaluations. Participants reported hearing more negative and fewer positive phrases from their caregivers than they reported using towards their own children. Similarities between the frequency of negative or positive phrases heard during participants’ upbringing and the frequency of phrases from those categories used with their own children could not be found. Neither parental warmth and control nor exposure to self-esteem-related phrases were significantly associated with self-esteem. Age differences in experienced parenting phrases and parenting phrase usage were observed, as younger participants reported hearing more positive phrases and using fewer negative phrases than older participants. In Study 2, the ratings of the parenting phrases were related to participants’ age and experienced parenting during childhood, but no association was found with parental status. Exploratory results from both studies align with the Relational Framing Theory, suggesting it may serve as a useful addition to traditional parenting style theories in interpreting caregiver-child communication. These studies provide a foundation for investigating parenting phrases, thereby offering a novel perspective on parental communication – a crucial factor in shaping child development and family dynamics.