Biomarker-anchored screening of dietary microplastics via a gut microbiota-informed machine learning model.

Machine learning-based prediction of drug lactation risk: Bridging molecular features and breastfeeding safety.

Comparative Analysis of Risk Stratifications for Non-Muscle-Invasive Bladder Cancer According to the Classifications Provided by the National Comprehensive Cancer Network, European Association of Urology, and Japanese Urological Association Guidelines.

Singapore's pragmatic governance of sex work, combining criminalisation and toleration, is exemplified by the Medical Surveillance Scheme (MSS) introduced in 1976 to manage sexually transmitted infections (STIs)/HIV among registered sex-workers. While the MSS provides conditional healthcare access, it excludes undocumented migrant sex-workers, leaving them vulnerable to health inequalities, precarity, and state violence. Drawing on an ethnographically-informed study conducted between March 2022 and July 2023 in collaboration with Singapore's sex-worker community, this study examines how migrant sex-worker bodies are regulated and marginalised under entangled logics of biopolitical inclusion (via medical surveillance) and sovereign exclusion (via consistent policing and moralisation). We argue that public health governance of migrant sex-workers functions not only for disease control but also as a mechanism of social ordering, sustaining the structure of violence tied to sexuality, mobility, and nation-building. By centring migrant sex-workers' lived experiences, we challenge dominant epidemiological framings that reduce them to "risk categories," instead unpacking how state interventions enact coercion, fear and danger. The study calls for a reorientation of public health policy from punitive regulation towards grounding in dignity, equality and justice.

A novel physics-informed AI framework for the assessment and prediction of indoor radon concentration and risk classification.

Beyond prediction: Advancing risk assessment and addressing the evidence deficit in the statutory child maltreatment context.

Myelodysplastic syndromes/neoplasms (MDS) constitute a very heterogeneous group of clonal myeloid neoplasms characterized by a variable clinical course and recurrent genetic abnormalities. Their treatment relies on risk classification as lower or higher-risk categories by the original International Prognostic Scoring System (IPSS) or the current revised (IPSS-R) or molecular IPSS-M. Higher-risk MDS (HR-MDS) are clonal hematopoietic disorders characterized by significant cytopenias, dysplastic changes, and a high propensity for progression to acute myeloid leukemia (AML). Up to 40% of them usually progress to AML within two years of diagnosis. Allogeneic stem cell transplantation (HSCT) remains the only potential cure and standard of care for eligible patients. Despite standard treatments such as Allo-HSCT and hypomethylating agents (HMAs), outcomes remain suboptimal. Recent advances have led to the development of novel therapeutic strategies, such as BCL-2 inhibitors (venetoclax), IDH1/2 inhibitors (ivosidenib, enasidenib), CD47 inhibitors (magrolimab), TIM-3 inhibitors (sabatolimab), XPO1 inhibitors (eltanexor), NEDD8-activating enzyme inhibitors (pevonedistat), TP53-targeted agents (eprenetapopt), liposomal chemotherapy (CPX-351), and oral HMA formulations. Combinations of hypomethylating agents with these new drugs, as first-line treatment, have to date not proven more efficacious than HMA monotherapy. This review summarizes the current therapeutic landscape on novel therapies for HR-MDS, highlighting their mechanism of action, efficacy, and demonstrates the unmet clinical need for more effective therapies.

Assessing Tumor Morphological Complexity Using Fractal Analysis of Contrast-Enhanced CT for Risk Stratification in Pediatric Neuroblastoma.

The World Health Organization (WHO) classification for pancreaticobiliary cytopathology stratifies risk of malignancy (ROM) and outcome for pancreatic fine-needle aspirations (FNAs). ROM for biliary cytology, particularly for WHO categories Pan-low (IV) and Pan-high (V), has not been well examined. Retrospective data of biliary cytology diagnosed from 2016 to 2019 were collected. WHO classification was assigned. Malignant outcome was determined by histologic correlation or clinical progression/death. Excluding clinical follow-up <3 months (if no histology), the ROM, odds ratio, performance characteristics, and overall survival were calculated. A total of 1056 biliary cytology cases (from 789 patients) were included. Sample types were: 757 (72%) biliary brush/wash, 115 (11%) FNA, and 184 (17%) other. Same-site histologic correlation was 292 (28%) and histology of metastasis in 124 (12%). The median follow-up (for benign) was 1431 days. Malignant outcome occurred in 704 (n=789, 89%) of patients. Absolute ROM (%) for WHO I to VII by biliary brush/wash and FNA were: 92, 72, 87, 0, 100, 94, 100, and 50, 46, 73, N/A, 100, 80, 100, respectively. A high-tier (V-VII) category had significantly higher odds of malignancy regardless of biopsy modality, presence of stent, or autoimmune cholangitis. Absolute ROM for biliary lesions is high for WHO I through III and is at least partly because of difficulty in sampling cancers, particularly those extrinsically obstructing the bile duct. The rarity of low-grade biliary lesions impedes AROM assessment for Pan-low (WHO IV). The clinical value of biliary Pan-high (WHO V) remains uncertain because of rarity of high-grade noninvasive lesions and difficulty in cytology diagnosis.

Study on quantitative assessment of urban gas pipeline leakage risk using a TFN-BN hybrid model

Anti-JC virus (JCV) antibody testing is central to progressive multifocal leukoencephalopathy (PML) risk stratification in patients treated with natalizumab. While STRATIFY JCV™ is the historical reference assay, IMMUNOWELL™ is increasingly used in the context of biosimilar adoption. We aimed to determine whether inter-assay discordance reflects true biological divergence or threshold-dependent interpretative effects within clinically relevant PML risk categories. In this single-center observational study, 250 consecutive patients with relapsing multiple sclerosis (RMS) underwent same-day paired STRATIFY and IMMUNOWELL testing. Agreement was evaluated under two interpretative frameworks: (i) manufacturer-defined analytical cutoffs and (ii) clinically adopted PML risk-based thresholds. Quantitative comparability was assessed using Pearson correlation and Bland-Altman analysis. Using analytical cutoffs, concordance was 78.0% (κ = 0.55), with discordance predominantly in borderline index ranges. When applying clinically adopted risk thresholds, concordance increased to 93.2% (κ = 0.73). Index values correlated strongly (r = 0.922; p < 0.0001). Bland-Altman analysis showed minimal mean bias (+0.06), with wider deviations confined largely to low-to-intermediate index values rather than high-risk profiles. Inter-assay discordance between STRATIFY and IMMUNOWELL is largely threshold-driven and substantially attenuated when applying clinically meaningful PML risk categories. These findings support practical interchangeability for routine monitoring in the biosimilar era, while underscoring the need for cautious interpretation of borderline results.

Ten-year risk prediction models for atherosclerotic cardiovascular disease in Chinese patients with schizophrenia-spectrum disorders: A population-based cohort study.

A prospective evaluation of attitudes, behaviors, and clinical outcomes among Jordanian children and adolescents with type 1 diabetes who fasted during Ramadan: a comparison of the 2021 and 2026 IDF-DAR risk scores.

Pathologic Upgrading in Biopsy Grade Group 1, PI-RADS ≥ 3 Prostate Cancer: Factors Associated With Clinically Significant Disease on Radical Prostatectomy.

Cardiovascular risk factors in adolescence, often persist into adulthood, with low cardiorespiratory fitness representing one of the strongest predictors of future cardiovascular disease. We aimed to classify cardiovascular risk associated with low cardiorespiratory fitness in adolescents using machine learning models based on anthropometric and muscular fitness variables. A cross-sectional dataset of 7,852 adolescents aged 13-16 years was analysed. Predictors included body mass index, waist circumference, waist-to-height ratio, standing long jump, push-ups, and sit-ups, while the binary outcome was cardiovascular risk derived from maximal oxygen consumption (VO2 max) cut-points. Data were standardised, class imbalance was addressed using Synthetic Minority Over-sampling TEchnique (SMOTE),and eight supervised classifiers were trained with stratified five-fold cross-validation and grid search. Ensemble tree-based methods outperformed kernel-based models. Gradient Boosting achieved the best balance between predictive performance (area under the curve-receiver operating curve [AUC-ROC] 0.716, F1-score 0.857, recall 0.760, accuracy 0.755), followed by Random Forest. SHapley Additive exPlanations (SHAP) analyses indicated that muscular fitness measures, particularly push-ups, sit-ups and standing long jump, contributed most to risk classification, whereas anthropometric indicators showed lower importance. These findings suggest that machine learning models built from school-based muscular fitness tests and basic anthropometry can discriminate adolescents at cardiovascular risk due to low cardiorespiratory fitness, offering a feasible, low-cost strategy to support early identification and targeted physical activity interventions.

Although areal bone mineral density (aBMD) measured by DXA is a good predictor of fractures, nearly half of low-trauma fractures occur in individuals without osteoporosis (T-score>-2.5 SD). This study investigated whether femoral strength estimated from 3D DXA-based finite element (FE) analysis enhances the prediction of incident fractures compared with conventional 2D DXA-derived aBMD. Baseline hip DXA scans from 740 postmenopausal women in the Geneva Retirees Cohort were analyzed. 3D reconstructions of the proximal femur were generated to estimate femoral strength using FE analysis and to derive structural bone parameters. Over a mean follow-up of 5.7±1.5years, 100 low-trauma fractures were recorded, including 44 major osteoporotic fractures (MOF), 89% at non-hip sites. Femoral strength, total and trabecular vBMD, some cortical parameters, and conventional areal BMD were all significantly associated with incident low-trauma fractures (risk increase of 23-38% per SD decrease) and major osteoporotic fractures (36-59%), with femoral strength showing hazard ratios (95% confidence interval) of 1.32 (1.08-1.60) and 1.50 (1.12-2.01), respectively. A strength threshold of 2600N improved fracture risk classification beyond conventional aBMD thresholds and outperformed the previously proposed fragile strength 3000N threshold. Notably, 25% of women with low-trauma fractures and 27% with MOF were non-osteoporotic by aBMD but exhibited fragile femoral strength (≤3000N), with 14% and 16% respectively below the very fragile threshold (≤2600N). In conclusion, 3D DXA-derived femoral strength provides complementary value to aBMD in identifying women at elevated fracture risk, particularly those not classified as osteoporotic by conventional criteria.

Pregnancy complications still constitute an enormous threat to both mothers and babies around the world. According to WHO estimates, about 712 women die daily from pregnancy and delivery-related complications with more than 90% of cases occurring in low- and lower-middle-income countries [1]. Conventional methods of assessing maternal and fetal health risk depend on human interpretation of clinical measurements and Cardiotocography (CTG) signals, which can be tedious and timeconsuming while being prone to significant inter-rater variability [2]. In this work, we propose a combined clinical decision support system called MaternaInsight which involves machine-learning models for both maternal and fetal health risk prediction. Maternal health risk prediction model identifies three risk categories – Low, Mid, and High based on LightGBM classifier optimized with Optuna and rebalanced with SMOTETomek, resulting in 90.64% of test accuracy (macro-F1=90.62%). Fetal classification is accomplished using a Stacking Ensemble approach with a logistic regression model as the meta-classifier, and the model is trained using 39 CTG-derived features, yielding an accuracy of 94.13% and macro-F1 of 88.66% on test data. Both classifiers are trained on data following the strict sequence of Split-Scale-Balance to avoid leakage. They also include interpretability of individual predictions using SHAP, making them clinically explainable. The entire pipeline is encapsulated in a four-page Streamlit web app containing a Clinical Reference and Performance Dashboard.

The hERG safety assay is crucial for assessing the risk of delayed repolarization and QT interval prolongation in investigational drugs, yet a lack of consensus exists on classifying drugs into appropriate risk categories based on assay results. We propose a two-stage statistical framework for quantitative assessment of hERG safety assay classification for investigational products adhering to ICH E14 S7b Q&A Best Practice protocols. Our framework employs a fixed margin equivalence testing approach to evaluate concurrent positive controls against the positive control reference in Stage 1, followed by a logistic regression model based on a T-score to classify the investigational product into two risk categories (QT effect ≥ 10 msec or < 10 msec) in Stage 2. Applied to real-world data, our method demonstrates the ability to classify investigational products with an AUROC of at least 0.9, using a pre-specified probability margin. Additionally, our approach addresses lab-to-lab variability in hERG safety assay assessment, providing a standardized, quantitative method for evaluating hERG assay results that could improve consistency in drug risk assessment across the pharmaceutical industry.

To report long-term oncological outcomes of men with intermediate-risk prostate cancer managed with active surveillance at a single centre. We retrospectively analysed a prospectively maintained Manitoba Prostate Centre active surveillance database. Included were men with disease classed per National Comprehensive Cancer Network risk categories: very low-, low-, favourable intermediate-, and unfavourable intermediate-risk prostate cancer managed with surveillance (2004-2023). Active surveillance involved serial prostate-specific antigen testing, digital rectal exam, and confirmatory/surveillance biopsies. Outcomes were treatment-free survival, metastasis-free survival, prostate cancer-specific survival, overall survival, and biopsy grade progression. Kaplan-Meier estimated survival; Cox regression evaluated variables associated with outcomes. Among 561 men, the median age was 65 years (interquartile range 59-69) and the median follow-up was 4.4 years. Overall, 256 men (45.6%) transitioned to definitive treatment. Five-year treatment-free survival declined stepwise by risk group: 84.8% in very low-risk, 61.4% in low-risk, 50.0% in favourable intermediate-risk, and 21.4% in unfavourable intermediate-risk disease (P < 0.001). On multivariable analysis, older age, higher prostate-specific antigen density, and grade group ≥2 were independently associated with earlier treatment. Metastatic progression and prostate cancer-specific mortality were rare across the cohort, with higher event rates observed among patients with unfavourable intermediate-risk disease. In this real-world active surveillance cohort, the likelihood of remaining on surveillance differed substantially by baseline risk, with unfavourable intermediate-risk disease associated with earlier transition to treatment and higher risks of disease progression. Older age, higher prostate-specific antigen density, and grade group ≥2 were independently associated with discontinuation of active surveillance. Despite less intensive surveillance protocols, rates of metastatic progression and prostate cancer-specific mortality remained low, particularly among patients with low-risk and favourable intermediate-risk disease, supporting active surveillance as a viable management strategy in carefully selected men beyond low-risk disease.

Our study aimed to determine if chemotherapy administration based on the 21-gene Oncotype DX Recurrence Score (RS) in men with hormone receptor positive (HR+) Her2 negative (Her2-) lymph node positive(LN+) breast cancers (BC) impacted overall survival (OS). We conducted a retrospective cohort study on adult men and women with HR+ Her2-, 1-3 axillary LN + BC, with a valid oncotype DX RS assay, diagnosed between the years 2004-2020, using the National Cancer Database. RS risk categories were defined as low risk: 0-13, intermediate risk:14-25, and high risk: ≥26. Of the 77,820 patients included in the study, 900 (1.2%) were male and 76,920 (98.8%) were female. Higher RS (both as a continuous and categorical variable) was significantly associated with worse OS in males (p = 0.003 continuous, p = 0.006 categorical), females ≤ 50 years old and females > 50 years old (p < 0.001 both continuous and categorical, in both groups). In the stratified adjusted models, there was no association between receipt of chemotherapy and OS in males for all RS risk groups. Conversely, chemotherapy improved OS for women ≤ 50 and for women > 50 who belonged to the intermediate and high RS risk groups (p = 0.001 and p < 0.001 respectively (women ≤ 50); p = 0.005 and p < 0·001respectively (women > 50)). RS as determined by the Oncotype DX assay is associated with OS in men and women with HR+ Her2- LN+ BC. However, RS is not an indicator of the benefit of chemotherapy on OS in men, in contrast to women, with HR+ Her2- LN+ BC.