Network intrusion classification referred to the process of monitoring and analyzing network traffic to identify suspicious activities or attacks. In this work, author proposed a novel approach to classify network intrusion by utilizing deep reinforcement learning (DRL), integrating a reservoir computing approach Echo State Network (ESN). A DRL-based approach improved upon traditional deep learning by adapting dynamically to novel/unknown and evolving attack patterns. Unlike static models, DRL continuously learned optimal strategies through interaction with the environment, allowing for better detection of previously unseen threats in real-time. To address the class imbalance often encountered in network intrusion datasets, we evaluated the performance of several advanced data balancing techniques, including Borderline-SMOTE, SMOTE-ENN, ADYSN, and K-means SMOTE. The findings demonstrated that the K-means-based data balancing method outperformed other techniques, resulting in the most robust performance across various metrics. Author conducted multi-dataset validation on benchmark datasets like NF-BoT-IoT, NF-UNSW-NB15, NF-ToN-IoT, NF-ToN-IoT-v2, NF-CSE-CIC-IDS2018 and NF-UNSW-NB15-v3 to ensure robustness across different network flow data. For adaptive modeling testing, author excluded some attack types from training data and included them in testing data (e.g., DoS, Backdoor attacks were excluded from the training data but included in the testing data (see Table 3)). The proposed approach enhanced the accuracy and reliability of intrusion detection, making it a viable solution for securing modern network infrastructures. The source code of this work is available in this Github repository (https://github.com/codewithkhurshed/DRLZDNIDS).

Worldwide, most refugees and asylum seekers are hosted in low- and middle-income countries, often with insufficient resources for public health surveillance. Refugees and asylum seekers living in camps may be particularly susceptible to infectious diseases because of the overcrowding and inadequate environmental health conditions inherent in camps. We propose that broader implementation of wastewater and environmental surveillance provides possibilities to support the health of refugees, asylum seekers, and the communities that host them by enhancing infectious disease surveillance. Key considerations include overcoming operational challenges, avoiding additional stigma, ensuring governance structures are regularly updated, and ceasing surveillance if data are weaponized or actionable data are not produced. (Am J Public Health. Published online ahead of print April 16, 2026:e1-e5. https://doi.org/10.2105/AJPH.2026.308433).

Fragment-Based Drug Discovery (FBDD) is a powerful strategy with a proven track record of generating potent bioactive small molecules from low-affinity chemical fragments. Computational approaches to FBDD are often limited by the availability of high-quality, structurally resolved data on fragment binding poses. To address this gap, we introduce the Structurally Augmented Fragment Repository (SAFR), a novel data set designed to support in silico FBDD. Initially, a set of 89,375 high-confident binding poses of bioactive molecules in public sources was obtained by applying a filtering protocol involving 2D ligand similarity and 3D ligand superposition against protein-bound ligand structures followed by scoring with protein-ligand docking and interaction features. Fragmentation of the bioactive ligands in their predicted binding poses resulted in a total of 818,385 fragment-protein interactions between 157,080 unique chemical fragments and environments from 1,142 distinct proteins. Of them, 270,155 are unique fragment-protein interactions, of which 237,284 (88%) are not represented in protein-bound ligands in the PDB. Case studies using SAFR for bioisosteric replacements and scaffold hopping are presented. SAFR is a useful resource to support fragment screening campaigns and hit-to-lead optimization. It is publicly available at https://zenodo.org/records/18229523.

Objectives. To investigate the association between layoff experience over 33 peak working years and subsequent premature death. Methods. This open cohort study involved data from 7234 working-age adults who took part in the US National Longitudinal Survey of Youth 1979. Cumulative layoff experience from the ages of 16 to 55 years was operationalized as the number of jobs ending in a layoff from 1979 to 2012. Mortality outcomes were assessed from 2012 to 2022, when the participants were younger than 65 years. Results. Layoff experience was associated with 18.71 (95% confidence interval [CI] = 1.09, 36.23; 1 layoff vs no layoffs) and 32.61 (95% CI = 2.66, 62.56; 2 or more layoffs vs no layoffs) excess deaths per 10 000 person-years. In a fully adjusted Cox proportional hazards model, experiencing 1 layoff and 2 or more layoffs (vs no layoffs) was associated with 1.23 (95% CI = 1.00, 1.51) and 1.30 times (95% CI = 0.96, 1.77) higher hazards of premature death from 2012 to 2022 (P trend = .02). The observed association did not statistically differ by timing of layoffs, sex/gender, or race/ethnicity. Conclusions. Layoffs may be an important determinant of mortality among US workers. (Am J Public Health. Published online ahead of print April 16, 2026:e1-e9. https://doi.org/10.2105/AJPH.2026.308432).

Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1), encoded by the hypervariable var gene family, is central to malaria pathogenesis, influencing both disease severity and immune evasion. Classifying var genes into upstream groups (upsA, upsB, upsC, upsE) is important for understanding parasite biology and clinical outcomes, but remains challenging, especially with partial sequences, such as the DBLα tag or RNA-Seq assemblies. We developed upsML, a machine-learning-based classifier trained on 2,530 curated var genes, to accurately assign upstream groups based on sequence features from different partial gene regions. We compared seven methods, including support vector machines, random forests, XGBoost, and HMMER models. Several models in upsML achieve accuracies of 83% for DBLα-tag sequences and 92% for full-length PfEMP1 sequences, thereby significantly outperforming existing tools. Additionally, we developed a model to distinguish internal from subtelomeric var genes, which we applied to a global collection of P. falciparum genomes, revealing a higher frequency of internal var genes in Asia. upsML is available at https://github.com/sii-scRNA-Seq/upsML, providing a robust and efficient resource for large-scale var gene analysis. It can classify var genes from 20 genomes in under one second.

Objectives. To quantify temporal changes in mental health and substance use emergency department utilization following Hurricane Helene (September 2024) in Western North Carolina and identify disparities across vulnerable populations. Methods. We conducted a cross-sectional analysis using emergency department surveillance data from Western North Carolina counties. We calculated incidence rate ratios (IRRs) with 95% confidence intervals (CIs) by comparing 3 recovery periods (immediate, early, short-term) with their corresponding calendar dates in 2023. Results. Alcohol-related visits increased across all periods (immediate: IRR = 1.35; 95% CI = 1.14, 1.60; early: IRR = 1.47; 95% CI = 1.31, 1.65; short-term: IRR = 1.36; 95% CI = 1.21, 1.52). Anxiety disorders increased across all periods, mood disorders increased during early recovery, schizophrenia visits increased during short-term recovery (IRR = 1.19; 95% CI = 1.02, 1.40), and opioid-related visits showed marginally significant increases during short-term recovery (IRR = 1.21; 95% CI = 1.00, 1.45). Counties with higher elderly populations showed amplified alcohol and anxiety effects; counties with higher uninsured or covered by Medicaid rates showed lower utilization. Conclusions. Hurricane Helene was associated with sustained increases in alcohol- and anxiety-related emergency department visits. Disparities across age and insurance status highlight differential vulnerability and access patterns requiring targeted interventions. (Am J Public Health. Published online ahead of print April 16, 2026:e1-e4. https://doi.org/10.2105/AJPH.2026.308456).

Current clinical tools for predicting relapse and guiding adjuvant therapy in clear cell renal cell carcinoma (ccRCC) lack precision, especially in intermediate-risk disease. This study evaluated whether a non-proliferative tumour cell phenotype, P21 (CDKN1A inhibitor) positive and MCM2 (DNA replication protein) negative (P21⁺/MCM2⁻), could serve as a robust biomarker to improve prognostic stratification and guide post-nephrectomy treatment decisions. We used multiplex immunofluorescence and AI-based image analysis on nephrectomy specimens from three independent ccRCC cohorts: UK arm of the SORCE trial (n=382), Korean (n=71), and Scottish (n=88). An optimal 2% cut-off for P21⁺/MCM2⁻ cells was determined using X-tile software. Additional analyses assessed endoglin/CD105 co-expression, paired primary-metastatic samples (n=41), and associations with adjuvant sorafenib therapy. In two intermediate-risk ccRCC cohorts (SORCE, n=63; Korean, n=71), patients with >2% P21⁺/MCM2⁻ cells had significantly longer time to relapse (HR=0.17, 95% CI: 0.06-0.54; HR=0.27, 95% CI: 0.10-0.72). Prognostic value was confirmed in high-risk (SORCE, HR=0.43, 95% CI: 0.19-0.99) and all-risk (Scottish, HR=0.37, 95% CI: 0.14-0.98) cohorts. Notably, patients with high P21⁺/MCM2⁻ levels on placebo fared better than those receiving adjuvant TKI therapy (HR=0.29, 95% CI: 0.16-0.50). In 41 paired samples, 85% showed higher P21⁺/MCM2⁻ abundance in metastases than in primary tumours. As a conclusion, P21⁺/MCM2⁻ cell count is a robust biomarker that refines relapse risk stratification in ccRCC and identifies patients who may not benefit from adjuvant tyrosine kinase inhibitor therapy. High levels of these non-proliferative, senescent-like cells suggest tumour dormancy and a more favourable outcome without treatment.

This study examines cross-linguistic influence (CLI) in the production of wh-questions by child heritage speakers of Mandarin aged 6;9 to 16;2 years in Canada. The form of CLI addressed is wh-fronting in Mandarin object-questions (where fronting is ungrammatical) and when-questions (where fronting is dispreferred). Our goals were, first, to determine whether the children front more frequently relative to their mothers; and second, to determine the impact of structural overlap between languages (operationalized by question type), relative language dominance, and age on the rate of children's fronting. Results show that the children front more frequently, and in more contexts, than do their mothers, indicating CLI. Structural overlap increased the likelihood of fronting for only some children, whereas greater dominance in English increased the likelihood of fronting for the group. Age did not correlate with changes in rates of fronting, indicating that CLI may sometimes reflect permanent divergence from the parental grammar.

Objectives. To provide a descriptive assessment of self-reported injury and life disruption attributable to the January 2025 Los Angeles County (LAC), California, firestorm among a sample of people experiencing homelessness (PEH). Methods. We integrated firestorm impact measures into an ongoing longitudinal survey of PEH in LAC. Adult PEH who completed the December 2024 and January 2025 surveys were included. Results. Among 374 respondents (average age 40 years, 45.7% female, 35.8% Hispanic, 23.5% non-Hispanic Black, 30.2% non-Hispanic White, 84.1% chronically homeless), 286 (76.5%) reported life disruption or injury from the firestorm. Compared with sheltered or housed respondents (n = 144), unsheltered vehicular respondents (n = 111) and unsheltered public respondents (n = 119) were more likely to report having their lives in danger, being injured, damage to living space, having to evacuate, damage or loss of belongings, and difficulty finding shelter. Those in evacuation zones (n = 48) more frequently reported having to evacuate, prolonged smoke exposure, and difficulty finding shelter. Conclusions. The firestorm caused injury or disruption for most survey participants, especially unsheltered PEH. Public Health Implications. Wildfire disaster preparedness and response should integrate homeless service coordination, with attention to unsheltered PEH. (Am J Public Health. Published online ahead of print April 9, 2026:e1-e8. https://doi.org/10.2105/AJPH.2026.308510).