This study aims to analyze the content of SITABAH messages, measure community digital literacy, and develop and conduct a limited trial of adaptive message prototypes to improve disaster communication effectiveness. A mixed-methods approach was applied, consisting of content analysis of 10 SITABAH messages from disaster events in January 2024, a digital literacy survey of 150 respondents across six sub-districts, in-depth interviews with system managers and community members, and a prototype trial with 50 respondents. Data were analyzed using descriptive quantitative and thematic qualitative techniques with triangulation. The findings indicate that SITABAH has not functioned optimally as an early warning system: no messages contained early warnings, most focused on victims and damage, instructions were general, and no visual elements were used. The digital literacy survey revealed relatively high levels of information access (64–80%), but lower levels of message comprehension (56–72%) and readiness to act (48–60%). Prototype trials demonstrated that visual and interactive formats were clearer and more effective in encouraging readiness to act compared to standard text. The study concludes that SITABAH’s effectiveness can be enhanced through the integration of early warning features, adaptive message design, improved digital literacy, and the development of a multi-channel SITABAH.

Asymmetric recovery of Euro-Mediterranean croplands driven by co-occurring climate forcings.

Caribbean Small Island Developing States (SIDS) are on the frontline of climate-induced coastal risks, where rising sea levels and intensifying storm surges converge with concentrated socioeconomic exposure. The inherent spatial constraints, economic dependency on tourism, and infrastructure clustering in coastal zones exacerbate systemic vulnerability. This study establishes an integrated high-resolution compound coastal risk framework by coupling hydrodynamic storm surge simulations with sea level projections and socioeconomic exposure data. The framework evaluates systemic risks across diverse return periods and Shared Socioeconomic Pathways including SSP1-1.9, SSP2-4.5, and SSP5-8.5, utilizing a modified static inundation model to integrate sea level rise and surge extremes for generating spatially explicit vulnerability assessments. Results revealed spatial heterogeneity and a distinctive ‘high-exposure–high-density–high-sensitivity’ configuration across several SIDS. Under SSP5–8.5, annual economic losses exceed USD 300 million in Jamaica and Cuba, and over 30,000 people may be affected in Dominica. Mitigation pathways consistent with a 1.5 °C warming limit (SSP1–1.9) reduce potential losses by 30–50%, while adaptation strategies, such as coastal ecosystem restoration and early-warning systems, deliver up to 40% carbon co-benefits. Beyond risk quantification, this framework supports climate-resilient infrastructure planning. By correlating spatial risks with adaptation finance, this research supports the Paris Agreement and Sustainable Development Goals (SDGs).

Identification of early hemodynamic changes ensures optimal patient outcomes. We examined if a second-generation, electronic health record-based pediatric hemodynamic software system with semi-automated early warning [Situational Awareness Vital Electronic Scout (SAVES)] identified actionable warning levels with greater accuracy. Using a retrospective medical records review, hospitalized nonintensive care children's encounters were included if they had first-generation warning system [Pediatric Early Warning System (PEWS)] software data. Clinicians manually entered data and warning levels were automatically created. An analyst applied PEWS data into the SAVES software that automatically included key patient characteristics from health records. Using descriptive analyses, PEWS and SAVES data were assessed in 3 ways: all data points, the first data point, and one randomly selected data point per encounter. In total, 693,962 PEWS data-point rows from 43,505 encounters among 26,131 unique in-patients were included. Of patients, 20% were transferred to intensive care during their hospital stay. Over 90% of the time points of data using PEWS and SAVES were equally accurate based on warning category level; however, in 6.0 to 8.86% of data points, the warning level varied, with SAVES prompting higher warning levels in all 3 assessment methods. Compared with the PEWS assessment scores, SAVES scores were 3.2 times (first record) to 13 times (all data points) more likely to have higher warning levels. The SAVES system identified higher early warning levels than PEWS. Accurate data, collected efficiently, facilitates recognition of subtle clinical changes before status deterioration.

Chicken disease detection and localization using multi-noise separation and acoustic recognition.

Transport and distribution patterns of floating marine litter: Numerical modeling and AI-empowered solutions.

Pediatric Early Warning Systems (PEWSs) have been used to predict adverse outcomes among hospitalized patients, though evidence is limited in resource-limited settings. We aimed to assess the performance of a novel PEWS score in a population of children with complicated severe acute malnutrition (SAM). Children hospitalized with SAM in Madarounfa, Niger were followed longitudinally from hospital admission to discharge. A novel Médecins Sans Frontières (MSF)-PEWS used scores for 9 clinical indicators, summarized into color categories (in increasing order of severity: green, orange, yellow, red) and assessed regularly throughout hospitalization. We estimated the association of color category at admission with subsequent clinical deterioration and mortality, as well as the association between time-varying color category with mortality within 48hours and time to discharge or death. In total, 4215 children were included in the analysis. Children in the red MSF-PEWS color category at admission had 2.58 (95% CI, 1.87-3.55) times greater risk of death during treatment compared with children in the green category at admission. In time-varying analyses, children classified in the red category at any given time were 11.9 (95% CI, 11.3-12.6) times more likely to die in the following 48hours than children in the green category. More severe color categories were associated with longer lengths of hospital stay and shorter times until death. MSF-PEWS scores were effective at identifying children at higher risk of adverse clinical outcomes, including mortality, during inpatient treatment for complicated SAM in Niger.

Non-destructive diagnostics and early-warning systems for Penicillium rot in fresh produce: A review of emerging technologies

The Cladonio–Pinetum association is among Europe’s most endangered dry, sandy forest communities and is highly sensitive to microclimate and edaphic shifts. We quantified how seasonality and microhabitat (psammophilous grasslands vs. pine forests) modulate physiological traits of the dominant terricolous lichens Cladonia mitis and Cl. uncialis in Bory Tucholskie National Park (north-central Poland). From September 2022 to September 2023 we monitored six permanent sites, recording light, temperature, moisture, and soil chemistry, and seasonally measuring maximum quantum efficiency of PS II ( F V / F M ) and concentrations of usnic acid, total chlorophyll ( a + b ), lutein, and β-carotene. Usnic acid was quantified by UHPLC–PDA; chlorophylls/carotenoids by UHPLC–ESI-QqQ-MRM. F V / F M peaked in winter (0.77 ± 0.019) and reached a summer minimum (0.59 ± 0.05) in both species, consistent with photothermal and drought stress. Usnic acid showed the opposite pattern, peaking in summer (31.66 mg g⁻¹ DW) and remaining ∼2 × higher in Cl. mitis across seasons; beyond certain concentration levels, higher usnic acid content was associated with a downward trend in F V / F M . Total chlorophyll differed between species and was consistently higher in Cl. mitis ; β-carotene showed seasonal variation, whereas lutein remained stable. Multiple regression identified light, air temperature, soil moisture, pH, and nitrogen as the main predictors of both F V / F M and usnic acid, explaining > 40% of their variance. Our results demonstrate that an integrated suite of chlorophyll fluorescence, pigment profiles, and secondary metabolite levels provides a sensitive early-warning system for detecting habitat degradation in Scots pine lichen forest. Managing canopy openness and curbing nutrient enrichment emerge as key conservation levers for sustaining these communities. • Photosystem II efficiency peaks in winter, dips in summer; microclimate–driven. • Usnic acid varies by season and lichen species. • Total chlorophyll is stable; β-carotene peaks in summer stress; lutein constant. • Physiological and environmental metrics are a tool to assess Cladonio-Pinetum state.

Extreme desert dust events and chronic PM2.5/PM10 exposure: A public health risk assessment in the Taklamakan region, China

Synergistic impacts of heat, pollen, and air pollution on allergic rhinitis and asthma under climate change: A 20-year time-series study.

Harmful cyanobacterial blooms (HABs) pose serious risks to freshwater ecosystems, drinking water supplies, and public health, highlighting the need for reliable early-warning systems. This study presents a rigorously validated machine learning framework for predicting cyanobacterial alert levels under strongly imbalanced conditions using routinely measured physicochemical variables. Four gradient boosting algorithms were systematically combined with 12 resampling strategies and evaluated within a nested cross-validation framework to ensure unbiased performance assessment. Model evaluation incorporated metrics tailored to imbalanced classification, including recall, F1-score, balanced accuracy (BA), and the Matthews correlation coefficient (MCC), with particular emphasis on the detection of alert events. Results demonstrate that resampling is critical for improving minority-class detection, with SMOTE-based approaches consistently providing the most favorable balance between sensitivity and precision across algorithms. LightGBM combined with SMOTE achieved the highest recall and F1-score, together with strong BA and MCC values and low variability across folds, indicating robust generalization. XGBoost combined with SMOTE exhibited a more balanced precision-recall profile with comparable overall performance but higher variability. SHAP-based interpretability analyses revealed consistent and ecologically meaningful drivers across models, with water temperature, turbidity, and pH emerging as the most influential predictors. By restricting inputs to variables measurable in near real time using low-cost insitu sensors, the proposed framework is designed to support operationally feasible early-warning applications through frequent updates of alert-level predictions within environmental monitoring systems. Overall, the findings highlight the importance of addressing class imbalance, ensuring rigorous validation, and incorporating interpretability to support practical and operationally feasible cyanobacterial early-warning applications.

Maximum mean discrepancy enhanced Informer for accurate cross-domain dual-timescale effluent prediction in wastewater treatment plants.

Pathogen surveillance in complex environmental matrices requires analytical methods that are sensitive, robust, and suitable for field deployment. Here, we report a geometry-engineered trimetallic PdPtRu nanozyme-enabled multimodal lateral flow immunoassay (LFIA) for pathogen analysis. The spiky porous architecture, together with multimetallic synergy, promotes enhanced photothermal conversion and catalytic signal transduction through localized charge redistribution and structural effects. Finite element simulations reveal an enhancement of the local electric field and an increased power dissipation density, thereby improving photothermal conversion efficiency. The intensified local field induces strong interfacial polarization, resulting in a 5.43-fold increase in surface ·O2- flux and a reduced reaction energy barrier for peroxide activation. When integrated into LFIAs, this label improved the detection sensitivity for Salmonella typhimurium by 200-fold compared with conventional colloidal gold assays. From sample to answer, we further developed a smartphone application embedding a convolutional neural network and coupled it with a portable 3D-printed signal acquisition module to construct a self-contained, intelligent, and field-deployable detection system. This platform achieves risk-level stratification with high predictive performance (R2 = 0.98, AUC = 0.99). This work establishes a scalable, modular foundation for early-warning surveillance systems, facilitating timely public health interventions.

The accelerating pace at which topics rise, peak, and fade across digital platforms has made early trend detection a critical capability for journalism, public health surveillance, marketing, and policy planning. Conventional trend-detection systems rely on single-platform signals and opaque deep learning models, which limits both their coverage and their trustworthiness in operational decision-making contexts. This paper proposes a multi-source social media analytics framework that fuses textual, temporal, and network signals from Twitter/X, Reddit, YouTube comments, and news RSS feeds to detect emerging trends substantially earlier than single-source baselines. The framework combines a burst-aware temporal feature extractor, a graph-based diffusion model capturing cross-platform propagation, and a gradient-boosted forecasting ensemble, with SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations) layers providing per-prediction interpretability. We evaluate the framework on a curated 14-month, multi-platform dataset spanning approximately 9.2 million posts across 1,800 manually and weakly labeled emerging topics. The proposed system achieves a mean detection lead time of 36.4 hours ahead of trend-platform native trending lists, an F1-score of 0.91 for trend/non-trend classification, and a mean absolute percentage error (MAPE) of 11.3% for 48-hour volume forecasting, outperforming LSTM, Prophet, and single-platform XGBoost baselines. Explainability analysis shows that burst acceleration, cross-platform co-occurrence, and influencer-adjusted reach are the most consistently influential features across topic categories. The results demonstrate that combining multi-source fusion with explainable machine learning yields both earlier and more trustworthy trend forecasts, with direct implications for misinformation early-warning systems, public health surveillance, and real-time marketing intelligence.

ABSTRACT Influenced by situational crime prevention (SCP) and zemiology, this research proposes situational harm reduction (SHR) as a new way of thinking about flood risk management (FRM). SCP adopts a problem-oriented approach to crime by implementing techniques and strategies that reduce opportunity structures. The SHR framework applies these principles towards flood adaptation, demonstrating how the development of bottom-up, complementary strategies of flood mitigation and resilience can help individual community members and businesses reduce the social and economic harms associated with flooding. Drawing on a qualitative research methodology that included semi-structured interviews and a focus group with a total of forty local residents and business owners in Matlock Town, Derbyshire, the findings show that members of the community used prior flooding experience, knowledge of flooding precipitators, and social networks to develop sustainable solutions in flood adaptation and resilience. The strategies adopted by individuals and business owners include property-level flood defenses, early-warning systems, individual drainage maintenance, recognition of the role of nature-based solutions in flood mitigation, and collective challenging of unsustainable urban developments. A SHR approach to FRM provides a holistic, multidisciplinary framework that may serve as a practical foundation for future collaboration between researchers, policymakers, and flood-affected communities alike.

ABSTRACT This study aims to explore the Early Warning System (EWS) with a particular focus on the needs and roles of women in the context of Bangladesh. Because of its geographic location, the country has been experiencing impacts of cyclones every year. While EWS in Bangladesh is well-developed, literature indicates that receiving and understanding early warnings by women compared to men is largely ignored. This empirical study was conducted at a cyclone-prone area located in the Southern coast of Bangladesh. The findings of this study show that the existing EWS is not yet gender sensitive. Women face difficulties in accessing and receiving early warning messages from formal channels. The messages they receive are often fragmented and therefore difficult to understand. This study also clearly explores that women face distinct challenges in accessing, receiving, interpreting, and understanding the early cyclone messages compared to men. Based on key findings, this study suggests that early warning for cyclones needs to be tailored according to the specific needs and roles of women.

ABSTRACT Air quality in Delhi has deteriorated significantly over the past decade, yet accurate high-resolution forecasting across multiple pollutants remains a major challenge due to heterogeneous monitoring networks, missing data, and complex spatial – temporal interactions. Motivated by the need for reliable early-warning systems, this study proposes DynLink-AQ, an end-to-end framework for multi-pollutant forecasting using data from 39 CPCB stations over 2009–2023. The system integrates rigorous data quality control and robust spatio-temporal imputation, followed by feature engineering enriched with meteorological drivers and temporal encodings. Unlike static distance-based station graphs, DynLink-AQ learns time-varying inter-station connectivity by inferring adaptive graph attention weights from station embeddings, spatial proximity, and temporal similarity, enabling event-driven and meteorology-linked coupling to be captured. Built upon this structure, the model alternates temporal attention blocks with spatial adaptive graph-attention layers to capture deep spatial – temporal dependencies. The framework supports multi-task prediction for 1–24-hour pollutant horizons with optional uncertainty quantification, and hyperparameters are tuned using the Enzyme Action Optimizer Algorithm (EAOA) under rolling-window training. Extensive walk-forward and spatial generalization experiments demonstrate strong predictive skill across pollutants, with ablation studies confirming the importance of dynamic connectivity learning and meteorological features. The study additionally provides GIS-ready outputs for seamless visualization and operational use in pollution management. Implications: DynLink-AQ enables operational, network-wide forecasting of PM2.5, PM10, NO2, and O3 across Delhi using Central Pollution Control Board monitoring data. By learning time-varying inter-station connectivity and combining spatial graph attention with temporal attention, the model improves 1–24 h predictions and provides uncertainty bounds for risk-aware alerts. Agencies can use these forecasts to issue timely hotspot formation, alerts, and plan short-term mitigation (traffic control, construction restrictions, industrial scheduling) during unexpected times. The framework is transferable to other cities with dense station networks and can integrate meteorological drivers already available hourly.

Airports are strategic targets for wastewater-based epidemiology because they concentrate highly mobile populations and may provide early signals of pathogen circulation. However, metagenomic investigations of airport wastewater remain limited, particularly in South America. Here, we present one of the first hybrid-capture target-enriched metagenomic investigations of airport wastewater in Brazil, integrating the detection of human-associated viruses and bacteriophage-derived host signatures to evaluate airports as sentinel surveillance sites. Seven untreated wastewater samples collected from a major Brazilian airport between December 2021 and March 2023 were concentrated, subjected to nucleic acid extraction, and analyzed using hybrid-capture target-enriched next-generation sequencing. Taxonomic analysis identified 615 viral and bacteriophage-associated taxa, including 440 viruses and 175 bacteriophages. Among the viral fraction, 21 human-associated viral taxa representing eight viral families were selected for detailed analysis. Norovirus GII was detected in all samples, while Mamastrovirus 1 and JC polyomavirus were detected in six of seven samples. SARS-CoV-2 and dengue virus type 1 were simultaneously detected in the March, 2023 sample. The bacteriophage fraction comprised 47 host-associated phage groups, with Streptococcus-associated phages predominating across samples. These findings demonstrate that airport wastewater can capture diverse human viral and bacteriophage-derived signatures associated with population mobility, supporting its application in environmental genomic surveillance and early-warning systems for emerging and circulating pathogens.

Impact of meteorological factors on other infectious diarrhea in mainland China: comprehensive risk assessment, forecasting and early warning.