Category: Statistics

Int J Hyg Environ Health. 2026 Jan 10;272:114744. doi: 10.1016/j.ijheh.2026.114744. Online ahead of print.

ABSTRACT

BACKGROUND: Prenatal exposure to tobacco smoke may impair neurodevelopment in children. However, accurately characterizing this exposure remains challenging.

METHODS: We pursued two objectives in this large population study. First, in 1708 pregnant women from the Environmental Influences on Child Health Outcomes (ECHO) cohort, we constructed Receiver Operating Characteristic (ROC) curves to determine urinary cotinine cut-offs to classify firsthand (FHS), environmental (ETS), and no exposure, and further distinguished secondhand (SHS) from thirdhand smoke (THS) exposure within ETS. Second, among 1593 participants in three pregnancy cohorts nested in ECHO, we fit multivariable linear regressions to examine the association between the newly defined smoke exposures and child full-scale intelligence quotient (IQ) at age 4-6 years, and to assess potential effect modification by maternal education or neighborhood deprivation.

RESULTS: Optimal cotinine cut-offs were 17.74 ng/mL and 0.44 ng/mL to discriminate FHS and no exposure, respectively. Among the ETS group, a cut-off of 5.69 ng/mL differentiated SHS from THS. Applying these optimal cut-offs, we estimated a 0.93-point (95 %CI: 3.44, 1.59) and a 1.03-point (95 %CI: 2.84, 0.79) lower child IQ in the FHS and ETS categories, respectively, compared to no exposure. The inverse association between prenatal ETS and child IQ was mainly driven by SHS. Stronger associations were suggested in subgroups with higher education attainment or those living in less deprived neighborhoods.

CONCLUSIONS: This study provides a novel classification of prenatal tobacco smoke exposures. Although the associations with child IQ were statistically insignificant, the study carries important implications for future research on developmental origins of diseases.

PMID:41520413 | DOI:10.1016/j.ijheh.2026.114744

Nutrition. 2025 Dec 15;144:113055. doi: 10.1016/j.nut.2025.113055. Online ahead of print.

ABSTRACT

OBJECTIVES: In this study we systematically assessed the diagnostic accuracy, calibration, and clinical utility of three study tools-the Nutritional Risk Screening 2002 (NRS2002), the Geriatric Nutritional Risk Index (GNRI), and the Patient-Generated Subjective Global Assessment Short Form (PG-SGA SF)-against the Global Leadership Initiative on Malnutrition (GLIM) criteria for identifying malnutrition in elderly patients with gastrointestinal malignancy. The aim was to determine their potential as pragmatic surrogates for the full GLIM diagnostic process.

METHODS: 412 patients (aged ≥ 60 y) with gastrointestinal malignancies from two hospitals in Shanghai were enrolled in this multicenter cross-sectional study. Diagnostic performance was assessed using GLIM criteria as the reference standard, evaluating the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Calibration was tested with the Hosmer-Lemeshow test, clinical net benefit was analyzed through decision curve analysis, and cross-center consistency was measured using the I² statistic.

RESULTS: The NRS2002 exhibited superior overall performance, characterized by high diagnostic accuracy (AUC = 0.85), the highest sensitivity (81%), excellent cross-center stability (I² = 0%), no significant calibration deviation (P = 0.415), and a clinical net benefit across a 0-96% risk threshold. The PG-SGA SF showed a comparable AUC (0.86), yet was accompanied by high specificity (87%), lower sensitivity (70%), significant calibration bias (P < 0.001), and notable inter-center heterogeneity (I² = 81.5%). The GNRI presented weaker diagnostic accuracy (AUC = 0.79) and significant calibration error (P = 0.039), though it maintained good cross-center stability (I² = 0%). All tools achieved an AUC > 0.70 across key clinical subgroups.

CONCLUSION: The NRS2002 is recommended as the primary surrogate diagnostic tool for GLIM-defined malnutrition in elderly patients with gastrointestinal malignancies, due to its balanced diagnostic accuracy and robust performance across settings. The GNRI offers an alternative based on objective parameters, while the PG-SGA SF is suitable for confirming malnutrition in low-risk outpatients. Future research should be focused on multicenter validation and examining the prognostic associations of these tools.

PMID:41520387 | DOI:10.1016/j.nut.2025.113055

Orv Hetil. 2026 Jan 11;167(2):43-50. doi: 10.1556/650.2026.33457. Print 2026 Jan 11.

ABSTRACT

INTRODUCTION: Acute ischemic stroke is one of the leading causes of disability and mortality in adults, representing a major public health burden also in Hungary. Functional recovery is influenced by a wide range of clinical, imaging and vascular risk factors; however, the relative contribution of these predictors is still poorly understood in the Hungarian population.

METHOD: The aim of this study was to identify the independent predictors of a 3-month functional outcome among ischemic stroke patients treated within 24 hours of symptom onset, using hierarchical linear regression based on data from the stroke registry of the National Laboratory of Translational Neuroscience.

RESULTS: A total of 1,062 patients were included in this retrospective cohort (mean age 71.9 ± 12.6 years, 52% female). The full model explained 20% of the variance (R² = 0.20, p<0.001). Age (β = -0.25; p<0.001), diabetes mellitus (β = -4.12; p = 0.02), chronic kidney disease (β = -5.45; p = 0.03), and multiple vascular disease (β = -3.89; p = 0.04) emerged as independent negative predictors. Among the acute clinical variables, admission NIHSS (β = -1.78; p<0.001) and ASPECTS (β = +2.34; p = 0.01) showed independent associations with favorable functional outcome, whereas intravenous thrombolysis administered within 60 minutes (β = +4.56; p = 0.02) was a positive predictor and beyond 60 minutes (β = -3.67; p = 0.04) a negative predictor; similarly, endovascular thrombectomy within 120 minutes (β = +5.23; p< 0.001) was a positive, and beyond 120 minutes (β = -4.89; p = 0.03) a negative predictor. Recanalization showed an independent positive effect (β = +0.12; p = 0.04).

CONCLUSION: Our findings indicate that functional recovery is primarily determined by stroke severity, the extent of ischemic injury, and patient age, whereas metabolic and vascular comorbidities exert a moderate but statistically significant negative effect on outcome. The separated time windows for intravenous thrombolysis and endovascular treatment underscore the importance of in-hospital treatment speed, reduce exposure misclassification, and enhance the explanatory power of the model. A key strength of the hierarchical approach is its ability to incorporate predictors in successive blocks, allowing for a more precise evaluation of the multifactorial determinants of functional recovery. Orv Hetil. 2026; 167(2): 43-50.

PMID:41520307 | DOI:10.1556/650.2026.33457

Indian J Med Res. 2025 Nov;162(5):639-643. doi: 10.25259/IJMR_1569_2025.

ABSTRACT

Background & objectives Pseudomonas aeruginosa is a leading cause of bacterial keratitis, known for its rapid progression, severe corneal damage, and resistance to treatment. Existing animal models exist to study disease mechanisms and therapeutic options require specialised conditions or complex procedures. This study aimed to develop a simplified, cost-effective, and reproducible murine model of P. aeruginosa keratitis using Swiss albino mice for translational research applications. Methods Four female Swiss albino mice (6-8 wk old) were maintained under standard conditions. Baseline ocular evaluation was done using a handheld slit lamp. The left corneas were abraded with a sterile 26G needle and inoculated topically with 10 μL of P. aeruginosa (1.0 × 10⁶ CFU/mL, ATCC 19660). The right eyes served as uninfected controls. Clinical signs were assessed on days 1, 2, and 3 using a standardised 0-4 scoring scale. Infection was confirmed through culture, biochemical tests, and PCR targeting the exotoxin A gene. Statistical analysis was performed using one-way ANOVA with Tukey’s post hoc test. Results All infected eyes developed progressive keratitis marked by lid swelling, corneal haze, and stromal involvement. Control eyes remained unaffected. Clinical scores increased significantly over time (P< 0.05). Culture and molecular analyses confirmed successful infection. Interpretation & conclusions This simplified Swiss albino mouse model effectively mimics human P. aeruginosa keratitis and enables cost-efficient study of pathogenesis and therapeutic testing. It can facilitate standardised ophthalmic infection research.

PMID:41520267 | DOI:10.25259/IJMR_1569_2025

Brief Bioinform. 2026 Jan 7;27(1):bbaf713. doi: 10.1093/bib/bbaf713.

ABSTRACT

RNA performs a variety of functions within cells and is implicated in various human diseases. Because druggable proteins occupy a small portion of the genome, considerable interest has been increasing in developing drugs targeting RNAs. Thus, precise prediction of small-molecule binding sites across different classes of RNAs is important. In this study, a lightweight deep learning program for predicting RNA-drug binding sites, called compound binding site prediction for RNA (CoBRA), is introduced. Our approach utilizes residue-level embeddings derived from a pre-trained RNA language model, without relying on any structural information. These embeddings encapsulate the contextual and statistical properties of each nucleotide and are used as input for a multi-layer perceptron classifier that performs binary classification of binding nucleotides. The model was trained using the TR60 and HARIBOSS datasets and tested on four independent benchmark sets. The performance of CoBRA demonstrates a relative improvement of 22.1% in the Matthew correlation coefficient and a 45.6% increase in sensitivity compared to existing state-of-the-art RNA-ligand binding site prediction methods that utilize structural information. These results demonstrate that sequence-based language model embeddings, which do not require explicit coordinate or distance information, can match or outperform structure-based methods. This makes it a flexible tool for predicting binding sites across diverse RNA targets.

PMID:41520231 | DOI:10.1093/bib/bbaf713

Brief Bioinform. 2026 Jan 7;27(1):bbaf711. doi: 10.1093/bib/bbaf711.

ABSTRACT

Statistical deconvolution methods offer a powerful solution for estimating cell-type-specific (CTS) profiles from readily available bulk tissue data. However, a critical limitation of existing methods is that they require the knowledge of cell type proportions of individuals in the bulk data. While the ground truth of cell type proportions in bulk samples are unknown, those methods use the estimated proportions to approximate the truth, which potentially introduces additional uncertainties in the inferred CTS profiles. To address this challenge, we propose Uncertainty-aware Bayesian Deconvolution (UBD) to incorporate uncertainty in cell type proportion estimates. By explicitly modeling the uncertainty in the initial estimates, UBD refines cell type proportions and estimates sample-level CTS data simultaneously. We show that UBD can improve the estimates of CTS profiles through extensive simulations. We further demonstrate the utility of UBD to reveal more CTS signals in its applications to two real datasets.

PMID:41520227 | DOI:10.1093/bib/bbaf711

Croat Med J. 2026 Jan 5;66(6):419-428.

ABSTRACT

AIM: To assess health care professionals’ knowledge and opinions regarding extracorporeal membrane oxygenation (ECMO) use, training, standards, and resource availability.

METHODS: This cross-sectional study employed an online self-administered survey to evaluate health care professionals’ knowledge and opinions concerning ECMO procedures. The survey consisted of multiple-choice and open-ended questions inquiring about demographics, ECMO practices, training and certification experiences, ECMO use during the COVID-19 pandemic, and obstacles to ECMO implementation.

RESULTS: The study enrolled 89 health care professionals from 12 countries. The respondents were most frequently from Kazakhstan (67.4%), Turkey (5.6%), Croatia (5.6%), and Ukraine (5.6%). Notably, 61.8% of respondents supported ECMO procedures performed by certified specialists. The respondents believed that the main ECMO indications were respiratory failure (83.1%), cardiopulmonary failure (69.6%), heart and lung transplantation (64.1%), and cardiogenic shock (58.4%). Major obstacles to ECMO implementation were reported to be high costs (53.9%), inadequately qualified staff (52.8% for physicians, 41.6% for nurses), and restricted availability of ECMO devices (42.7%).

CONCLUSION: The findings emphasize the need for standardized training, wider availability of ECMO standards, and efforts to address resource-related barriers to ECMO access. Our results primarily reflect practices in Kazakhstan and should be interpreted in light of the study’s restricted geographical coverage.

PMID:41520203

Croat Med J. 2026 Jan 5;66(6):390-398.

ABSTRACT

AIM: To examine the associations between marital status, place of residence, self-reported health status, quality of life, and physical activity among older adults during the COVID-19 pandemic.

METHODS: This cross-sectional study enrolled 962 participants aged 65 and older, surveyed between March 2020 and May 2023. Respondents were categorized according to marital status (married/living with a partner, single, divorced, widowed) and place of residence (own home vs nursing home). Standardized instruments were used: the Short Form Health Survey-36 for health status, the Personal Well-being Index for quality of life, and the Croatian short version of the International Physical Activity Questionnaire.

RESULTS: Respondents who were married or living with a partner reported significantly higher levels of physical activity, better physical and mental health, and greater life satisfaction than single, divorced, or widowed respondents (P<0.001). Community-dwelling respondents scored significantly higher on most health and quality-of-life indicators than nursing home residents, except for perceived future security.

CONCLUSION: Marital status and living arrangements significantly affected the self-perceived health, physical activity, and quality of life of older adults during the COVID-19 pandemic. The results emphasize the importance of social support and residential context in promoting healthy aging.

PMID:41520200

Inflamm Bowel Dis. 2026 Jan 11:izaf313. doi: 10.1093/ibd/izaf313. Online ahead of print.

NO ABSTRACT

PMID:41520196 | DOI:10.1093/ibd/izaf313

J Am Med Inform Assoc. 2026 Jan 11:ocaf230. doi: 10.1093/jamia/ocaf230. Online ahead of print.

ABSTRACT

BACKGROUND: Despite rapid integration into clinical decision-making, clinical large language models (LLMs) face substantial translational barriers due to insufficient structural characterization and limited external validation.

OBJECTIVE: We systematically map the clinical LLM research landscape to identify key structural patterns influencing their readiness for real-world clinical deployment.

METHODS: We identified 73 clinical LLM studies published between January 2020 and March 2025 using a structured evidence-mapping approach. To ensure transparency and reproducibility in study selection, we followed key principles from the PRISMA 2020 framework. Each study was categorized by clinical task, base architecture, alignment strategy, data type, language, study design, validation methods, and evaluation metrics.

RESULTS: Studies often addressed multiple early stage clinical tasks-question answering (56.2%), knowledge structuring (31.5%), and disease prediction (43.8%)-primarily using text data (52.1%) and English-language resources (80.8%). GPT models favored retrieval-augmented generation (43.8%), and LLaMA models consistently adopted multistage pretraining and fine-tuning strategies. Only 6.9% of studies included external validation, and prospective designs were observed in just 4.1% of cases, reflecting significant gaps in translational reliability. Evaluations were predominantly quantitative only (79.5%), though qualitative and mixed-method approaches are increasingly recognized for assessing clinical usability and trustworthiness.

CONCLUSION: Clinical LLM research remains exploratory, marked by limited generalizability across languages, data types, and clinical environments. To bridge this gap, future studies must prioritize multilingual and multimodal training, prospective study designs with rigorous external validation, and hybrid evaluation frameworks combining quantitative performance with qualitative clinical usability metrics.

PMID:41520192 | DOI:10.1093/jamia/ocaf230