Search Results (12,451)

Background: Ward patients who experience clinical deterioration are at high risk of mortality. Conventional rapid response systems (RRS) using track-and-trigger protocols have not consistently demonstrated improved outcomes. This study evaluated the impact of an artificial intelligence (AI)-based cardiac arrest prediction model. Methods: This 1-year, prospective, non-randomized interventional trial assigned hospitalized patients with AI-based software as a medical device (AI-SaMD) high-risk alerts to groups based on their subsequent clinical response; those reassessed or treated within 24 h comprised the AI-SaMD-guided cohort, while the remainder formed the usual care cohort. Alerts prompted an optional but not mandatory treatment review. The primary outcome was ward-based cardiac arrest; the secondary outcome was in-hospital mortality. Multivariable regression analysis was used to adjust for potential confounders. Results: Of 35,627 general ward admissions, 2906 triggered an AI-SaMD alert. Among these, 1409 (48.4%) were allocated to the AI-SaMD-guided cohort. The incidence of cardiac arrest significantly decreased from 2.07% to 1.06% (adjusted risk ratio (RR), 0.54; 95% confidence interval (CI), 0.20–0.88; p < 0.01). In-hospital mortality also significantly declined (adjusted RR, 0.65; 95% CI, 0.32–0.98; p < 0.05). Conclusions: AI-SaMD-guided alerts were associated with reductions in cardiac arrest and in-hospital mortality without requiring additional resources, supporting their integration into current clinical workflows to improve patient safety and optimize RRS performance. Full article

Wireless Sensor Networks (WSNs) are pivotal for data acquisition, yet reliability is severely constrained by routing voids induced by sparsity, uneven energy, and high dynamicity. To address these challenges, the Hybrid Acoustic-Optical Adaptive Void-handling Protocol (HAO-AVP) is proposed to satisfy the requirements for highly reliable communication in complex underwater environments. First, targeting uneven energy, a reinforcement learning mechanism utilizing Gini coefficient and entropy is adopted. By optimizing energy distribution, voids are proactively avoided. Second, to address routing interruptions caused by the high dynamicity of topology, a collaborative mechanism for active prediction and real-time identification is constructed. Specifically, this mechanism integrates a Markov chain energy prediction model with on-demand hop discovery technology. Through this integration, precise anticipation and rapid localization of potential void risks are achieved. Finally, to recover damaged links at the minimum cost, a four-level progressive recovery strategy, comprising intra-medium adjustment, cross-medium hopping, path backtracking, and Autonomous Underwater Vehicle (AUV)-assisted recovery, is designed. This strategy is capable of adaptively selecting recovery measures based on the severity of the void. Simulation results demonstrate that, compared with existing mainstream protocols, the void identification rate of the proposed protocol is improved by approximately 7.6%, 8.4%, 13.8%, 19.5%, and 25.3%, respectively, and the void recovery rate is increased by approximately 4.3%, 9.6%, 12.0%, 18.4%, and 24.2%, respectively. In particular, enhanced robustness and a prolonged network life cycle are exhibited in sparse and dynamic networks. Full article

Non-communicable diseases (NCDs) impose an overwhelming burden on global health systems. Prevailing healthcare for NCDs remains largely hospital-centered, episodic, and reactive, rendering them poorly suited to address the long-term, heterogeneous, and multifactorial nature of NCDs. Rapid advances in digital technologies, artificial intelligence (AI), and precision medicine have catalyzed the development of an integrative framework for digital-intelligent precision health management, characterized by the functional integration of data, models, and decision support. It is best understood as an integrated health management framework operating across three interdependent dimensions. First, it is grounded in multidimensional health-related phenotyping, enabled by continuous digital sensing, wearable and ambient devices, and multi-omics profiling, which together allow for comprehensive, longitudinal characterization of individual health states in real-world settings. Second, it leverages intelligent risk warning and early diagnosis, whereby multimodal data are fused using advanced machine learning algorithms to generate dynamic risk prediction, detect early pathological deviations, and refine disease stratification beyond conventional static models. Third, it culminates in health management under intelligent decision-making, integrating digital twins and AI health agents to support personalized intervention planning, virtual simulation, adaptive optimization, and closed-loop management across the disease continuum. Framed in this way, digital-intelligent precision health management enables a fundamental shift from passive care towards proactive, anticipatory, and individual-centered health management. This Perspectives article synthesizes recent literature from the past three years, critically examines translational and ethical challenges, and outlines future directions for embedding this framework within population health and healthcare systems. Full article

Background/Objectives: Chronic Kidney Disease (CKD) is a global public health burden with a rising prevalence driven by population aging. Existing prediction models, such as the Kidney Failure Risk Equation (KFRE), often lack generalizability across ethnicities and comprehensive systemic indicators. This study aimed to develop and validate a machine learning model for predicting CKD progression by integrating traditional risk factors with novel composite indicators reflecting systemic health. Methods: Data from the China Health and Retirement Longitudinal Study (CHARLS, n = 2500) was used for model training. External validation was performed using independent cohorts from the English Longitudinal Study of Ageing (ELSA, n = 1200) and the Health and Retirement Study (HRS, n = 1500). Multiple machine learning algorithms, including XGBoost, were employed. Feature engineering incorporated composite indicators such as the frailty index (FI), triglyceride–glucose (TyG) index, and aggregate index of systemic inflammation (AISI). Results: The XGBoost model achieved an area under the curve (AUC) of 0.892 in the training set and maintained robust performance in external validation (AUC 0.867 in ELSA, 0.871 in HRS), outperforming the KFRE (AUC 0.745). SHAP analysis identified the FI as the most influential predictor. Decision curve analysis confirmed the model’s clinical utility. Conclusions: This machine learning model demonstrates high accuracy and cross-ethnicity validity, offering a practical tool for early intervention and personalized CKD management. Future work should address limitations such as the retrospective design and expand validation to underrepresented regions. Full article

The maritime sector is under growing pressure to decarbonize, driving the adoption of alternative fuels such as methane, methanol, ammonia, and hydrogen. This study evaluates their thermal behavior and associated risks using Engineering Equation Solve software for heat transfer modeling and Areal Locations of Hazardous Atmospheres software for dispersion and explosion analysis in pipelines and storage scenarios. Results indicate that methane presents moderate and predictable risks, mainly from thermal effects in fires or Boiling Liquid Expanding Vapor Explosion events, with low toxicity. Methanol offers the safest operational profile, stable at ambient temperature and easily manageable, though it remains slightly flammable even when diluted. Ammonia shows the greatest toxic hazard, with impact distances reaching several kilometers even when emergency shutoff systems are active. Hydrogen, meanwhile, poses the most severe flammability and explosion risks, capable of autoignition and generating destructive overpressures. Thermal analysis highlights that cryogenic fuels require complex insulation systems, increasing storage costs, while methanol and gaseous hydrogen remain thermally stable but have lower energy density. The study concludes that methanol is the most practical transition fuel, when comparing thermal behavior and associated risks, while hydrogen and ammonia demand further technological and regulatory development. Proper insulation, ventilation, and automatic shutoff systems are essential to ensure safe decarbonization in maritime transport. Full article

Background/Objectives: Critically ill adults in intensive care units (ICUs) remain at high risk of death, while commonly used severity scores are complex and not always available at admission. We evaluated peri-admission lactate-to-albumin (LAR), C-reactive protein-to-albumin (CAR) and procalcitonin-to-albumin (PAR) ratios at ICU entry to predict ICU mortality in a cardiovascularly burdened cohort. Methods: We performed a single-centre retrospective observational cohort study in a tertiary cardiac ICU including adult admissions in 2024 with complete peri-admission lactate, C-reactive protein, procalcitonin and albumin. Results: Of 212 ICU admissions, 137 met the inclusion criteria. ICU mortality was 48.9%. Non-survivors had higher composite ratios and lower albumin than survivors. In multivariable models, LAR and CAR, but not PAR, remained independently associated with ICU mortality after adjustment for age, sex, and admission category. Receiver operating characteristic areas under the curve (AUC) were 0.692 for LAR, 0.677 for CAR and 0.625 for PAR. Cut-offs of LAR ≥ 0.106, CAR ≥ 3.18 and PAR ≥ 0.143 identified high-risk subgroups, with odds ratios for death of 6.18, 4.20 and 2.70, respectively, compared with lower-ratio patients, and LAR provided the best overall discrimination. Conclusions: Peri-admission LAR, CAR and PAR derived from routine tests in the ICU are associated with ICU mortality in critically ill adults, with LAR and CAR providing independent prognostic information and LAR showing the best discrimination. These simple composite ratios may complement severity scores for early risk stratification and warrant external validation. Full article

Uncertainty-aware decision-making increasingly relies on multimodal sensing pipelines that must fuse correlated measurements, propagate uncertainty, and trigger reliable control actions. This study develops a unified mathematical framework for multimodal data fusion and Bayesian decision-making under uncertainty. The approach integrates adaptive Covariance Intersection (aCI) for correlation-robust sensor fusion, a Gaussian state–space backbone with Kalman filtering, heteroskedastic Bayesian regression with full posterior sampling via an affine-invariant MCMC sampler, and a Bayesian likelihood-ratio test (LRT) coupled to a risk-sensitive proportional–derivative (PD) control law. Theoretical guarantees are provided by bounding the state covariance under stability conditions, establishing convexity of the aCI weight optimization on the simplex, and deriving a Bayes-risk-optimal decision threshold for the LRT under symmetric Gaussian likelihoods. A proof-of-concept agro-environmental decision-support application is considered, where heterogeneous data streams (IoT soil sensors, meteorological stations, and drone-derived vegetation indices) are fused to generate early-warning alarms for crop stress and to adapt irrigation and fertilization inputs. The proposed pipeline reduces predictive variance and sharpens posterior credible intervals (up to 34% narrower 95% intervals and 44% lower NLL/Brier score under heteroskedastic modeling), while a Bayesian uncertainty-aware controller achieves 14.2% lower water usage and 35.5% fewer false stress alarms compared to a rule-based strategy. The framework is mathematically grounded yet domain-independent, providing a probabilistic pipeline that propagates uncertainty from raw multimodal data to operational control actions, and can be transferred beyond agriculture to robotics, signal processing, and environmental monitoring applications. Full article

Background: The persistently low 5-year survival rate for pancreatic cancer (PC) underscores the critical need for early detection. However, population-wide screening remains impractical. Artificial Intelligence (AI) models using electronic health record (EHR) data offer a promising avenue for pre-symptomatic risk stratification. Objective: To systematically review and meta-analyze the performance of AI models for PC prediction based exclusively on structured EHR data. Methods: We systematically searched PubMed, MedRxiv, BioRxiv, and Google Scholar (2010–2025). Inclusion criteria encompassed studies using EHR-derived data (excluding imaging/genomics), applying AI for PC prediction, reporting AUC, and including a non-cancer cohort. Two reviewers independently extracted data. Random-effects meta-analysis was performed for AUC, sensitivity (Se), and specificity (Sp) using R software version 4.5.1. Heterogeneity was assessed using I² statistics and publication bias was evaluated. Results: Of 946 screened records, 19 studies met the inclusion criteria. The pooled AUC across all models was 0.785 (95% CI: 0.759–0.810), indicating good overall discriminatory ability. Neural Network (NN) models demonstrated a statistically significantly higher pooled AUC (0.826) compared to Logistic Regression (LogReg, 0.799), Random Forests (RF, 0.762), and XGBoost (XGB, 0.779) (all p < 0.001). In analyses with sufficient data, models like Light Gradient Boosting (LGB) showed superior Se and Sp (99% and 98.7%, respectively) compared to NNs and LogReg, though based on limited studies. Meta-analysis of Se and Sp revealed extreme heterogeneity (I² ≥ 99.9%), and the positive predictive values (PPVs) reported across studies were consistently low (often < 1%), reflecting the challenge of screening a low-prevalence disease. Conclusions: AI models using EHR data show significant promise for early PC detection, with NNs achieving the highest pooled AUC. However, high heterogeneity and typically low PPV highlight the need for standardized methodologies and a targeted risk-stratification approach rather than general population screening. Future prospective validation and integration into clinical decision-support systems are essential. Full article

Near-infrared (NIR) spectroscopy is a vital non-destructive analytical tool in the food and aquaculture industries. This study pioneers the application of portable NIR spectrometers for evaluating selenium (Se) content in the Pacific oyster (Crassostrea gigas). We developed quantitative and qualitative models to predict selenium levels in oyster tissue, representing a novel application for monitoring trace elements in marine organisms. Quantitative models were developed using partial least squares (PLS) regression on spectra collected with two portable spectrometers (Micro NIR 1700, Micro PHAZIR RX) and a benchtop FT-NIR instrument, with validation via cross-validation and an independent set. Qualitative models were also constructed to categorize Se content into three levels: 0–1, 1–3, and >3 mg/kg. For quantitative analysis, the Micro NIR 1700 model performed robustly in external validation (RP = 0.932; RMSEP = 0.392; RPD = 2.46). The Micro PHAZIR RX model achieved the highest RC (0.988) and the lowest RMSEC (0.233), yet cross-validation indicated a potential risk of overfitting. In contrast, the FT-NIR instrument yielded the best external predictive ability for powdered samples (RP = 0.954, RPD = 2.60), highlighting its high precision under laboratory conditions. For qualitative discrimination, the Micro PHAZIR RX’s classification module achieved a 100% correct recognition rate (AUC = 0.937). The models based on the Micro NIR 1700 and FT-NIR instruments showed cumulative contribution rates (CCR) of 98.61% and 97.59%, respectively, with high performance indices (PI) of 89.3 and 90.2, confirming their effective discrimination capability. The models established in this study enable the rapid, on-site detection of Se content in oyster samples, underscoring the significant potential of portable NIR spectroscopy for selenium analysis in shellfish. Full article

Artificial intelligence (AI) has rapidly evolved from an experimental tool in spine research to a multi-domain framework that has significantly influenced imaging analysis, surgical decision-making, and individualized outcome prediction. Recent advances have expanded beyond isolated applications, enabling automated image interpretation, patient-specific risk stratification, discovery of qualitative phenotypes, and integration of heterogeneous clinical and biomechanical data. These developments signal a shift toward more comprehensive, context-aware analytic systems capable of supporting complex clinical workflows in spine care. Despite these gains, widespread clinical adoption remains limited. High internal performance metrics do not consistently translate into reliable generalizability, interpretability, or real-world clinical readiness. Persistent challenges, which include dataset heterogeneity, transportability across institutions, alignment with clinical decision-making processes, and appropriate validation strategies, continue to constrain widespread implementation. In this perspective, we synthesize post-2019 advances in spine AI across key application domains: imaging analysis, predictive modeling and decision support, qualitative phenotyping, and emerging hybrid and language-based frameworks through a unified clinical-readiness lens. By examining how methodological progress aligns with clinical context, validation rigor, and interpretability, we highlight both the transformative potential of AI in spine research and the critical steps required for responsible, effective integration into routine clinical practice. Full article

The increasingly frequent and severe natural disasters have posed significant challenges to the resilience of power systems worldwide, creating an urgent need to investigate the security issues associated with these extreme events and to develop effective risk mitigation strategies. Meanwhile, as one of the leading topics in current research, artificial intelligence (AI) has demonstrated outstanding performance across various domains, such as AI-driven smart grids and smart cities. In particular, its efficiency in processing big data and solving complex computational problems has made AI a powerful tool for supporting decision-making in complex scenarios. This article presents a focused overview of power system resilience against natural disasters, highlighting recent advancements in AI-based approaches aimed at enhancing system security and response capabilities. It begins by introducing various types of natural disasters and their corresponding impacts on power systems. Then, a systematic overview of AI applications in power systems under disaster scenarios is provided, with a classification based on the task categories, i.e., predictive, descriptive and prescriptive tasks. Following this, this article analyzes current research trends and finds a growing shift from knowledge-based models towards data-driven models. Furthermore, this paper discusses the major challenges in this research field, including data processing, data management, and data analytics; the challenges introduced by large language models in power systems; and the limitations related to AI model interpretability and generalization capability. Finally, this article outlines several potential future research directions. Full article

Background: Cardiovascular polypharmacy inherently amplifies the risk of drug–drug interactions (DDIs), yet most studies remain limited to isolated drug pairs or predefined high-risk classes, without mapping the systemic architecture through which interactions accumulate. Objectives: To characterize the burden, severity, and network structure of potential DDIs in a real-world cohort of hospitalized cardiovascular patients using interaction profiling combined with graph-theoretic network analysis. Methods: This retrospective observational study included 250 hospitalized cardiovascular patients. All home medications at admission were analyzed using the Drugs.com interaction database, and a drug interaction network was constructed to compute topological metrics (i.e., degree, betweenness, and eigenvector centrality). Results: Polypharmacy was highly prevalent, with a mean of 7.7 drugs per patient, and 98.4% of patients exhibited at least one potential DDI. A total of 4353 interactions were identified, of which 12.1% were classified as major, and 35.2% of patients presented high-risk profiles with ≥3 major interactions. Interaction burden showed a strong correlation with medication count (r = 0.929). Network analysis revealed a limited cluster of hub medications, particularly pantoprazole, furosemide, spironolactone, amiodarone, and perindopril, that disproportionately governed both interaction density and high-severity risk. Conclusions: These findings move beyond conventional pairwise screening by demonstrating how interaction risk propagates through interconnected therapeutic networks. The study supports the integration of hub-focused deprescribing, targeted monitoring strategies, and network-informed clinical decision support to mitigate DDI risk in cardiovascular polypharmacy. Future studies should link potential DDIs to clinical outcomes and validate network-based prediction models in prospective settings. Full article

Rainfall, as the main driving force of natural disasters such as floods and droughts, has strong non-linear and abrupt characteristics, which makes it difficult to predict. As extreme weather events occur frequently in the Yellow River Basin, it is especially critical to reveal the physical mechanism of rainfall in the basin and integrate monthly scale meteorological data to achieve monthly rainfall prediction. In this paper, we propose a rainfall prediction model coupled with a physical mechanism and a temporal convolutional network (TCN) to achieve the prediction of monthly rainfall in the basin, aiming to reveal the physical mechanism between rainfall factors in the basin based on the transfer entropy and the multidimensional Copula function and based on the physical mechanism which is embedded into the TCN to construct a dual-driven prediction model with both physical knowledge and data, while the SHAP is used to analyze the interpretability of the prediction model. The results are as follows: (1) Temperature, relative humidity, and evaporation are key characteristic factors driving rainfall. (2) The physical mechanism features between temperature, relative humidity, and evaporation can be described by the three-dimensional Gumbel–Hougaard Copula function, with a more concentrated data distribution of their joint distribution probability. (3) The PHY-TCN model can accurately fit the extremes of the rainfall series, improving the model accuracy in the training set by 3.82%, 1.39%, and 9.82% compared to TCN, CNN, and LSTM, respectively, and in the test set by 6.04%, 2.55%, and 8.91%, respectively. (4) Embedding physical mechanisms enhances the contribution of individual feature variables in the PHY-TCN model and increases the persuasiveness of the model. This study provides a new research framework for rainfall prediction in the YRB and analyzes the physical relationship between the input data and output results of the deep learning model. It has important practical significance and strategic value for guiding the optimal scheduling of water resources, improving the risk management level of the basin, and promoting the ecological protection and high-quality development of the YRB. Full article

Background/Objectives: Bail-out stenting remains a procedural challenge for percutaneous coronary intervention (PCI) performed with drug-coated balloons (DCBs). No dedicated bedside tool is currently available to predict this event. We aimed to develop and internally validate a bedside Bail-Out Risk Score. Methods: We analyzed patients treated with DCBs between 2021 and 2025. Predictors of bailout stenting were identified through univariate analysis, and variables with p < 0.10 were entered into a multivariable logistic regression model. Regression coefficients were then transformed into integer points using the Sullivan method. Model performance was evaluated by AUC-ROC, calibration, and bootstrap internal validation (B = 1000). Results: A total of 352 patients (399 de novo lesions) were treated with DCB-only PCI. Bail-out stenting occurred in 14.5% of lesions (58/399). Independent predictors of bail-out stenting were prior CABG (OR 4.29, p = 0.002), proximal lesion location (OR 2.99, p = 0.003), and diffuse disease (OR 2.18, p = 0.018). Prior PCI (OR 0.44, p = 0.009) and lipid-lowering therapy (OR 0.42, p = 0.029) were protective, while LAD involvement showed a non-significant trend (OR 1.57, p = 0.137). The model demonstrated moderate discrimination (AUC = 0.734; optimism-corrected AUC = 0.704) and excellent calibration (intercept = 0.000, slope = 1.000). The final score (range –4 to +8) stratified lesions into low (≤–1), intermediate (0–3), and high (≥3) risk groups, with progressively higher predicted probabilities (≤9%, 13–37%, and ≥49%). Conclusions: The Bail-Out Risk Score provides a practical and reliable bedside tool to estimate procedural risk during stentless PCI. Full article

This observation-based and explainable approach demonstrates the applicability of multi-source remote sensing for flood assessment in data-scarce regions, offering a robust scientific basis for flood management and spatial planning in monsoon-affected areas. Floods are among the most frequent and devastating natural hazards, particularly in developing countries such as Myanmar, where monsoon-driven rainfall and inadequate flood-control infrastructure exacerbate disaster impacts. This study presents a satellite-driven and interpretable framework for high-resolution flood susceptibility and risk assessment by integrating multi-source remote sensing and geospatial data with ensemble machine-learning models—Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM)—implemented on the Google Earth Engine (GEE) platform. Eleven satellite- and GIS-derived predictors were used, including the Digital Elevation Model (DEM), slope, curvature, precipitation frequency, the Normalized Difference Vegetation Index (NDVI), land-use type, and distance to rivers, to develop flood susceptibility models. The Jenks natural breaks method was applied to classify flood susceptibility into five categories across Myanmar. Both models achieved excellent predictive performance, with area under the receiver operating characteristic curve (AUC) values of 0.943 for XGBoost and 0.936 for LightGBM, effectively distinguishing flood-prone from non-prone areas. XGBoost estimated that 26.1% of Myanmar’s territory falls within medium- to high-susceptibility zones, while LightGBM yielded a similar estimate of 25.3%. High-susceptibility regions were concentrated in the Ayeyarwady Delta, Rakhine coastal plains, and the Yangon region. SHapley Additive exPlanations (SHAP) analysis identified precipitation frequency, NDVI, and DEM as dominant factors, highlighting the ability of satellite-observed environmental indicators to capture flood-relevant surface processes. To incorporate exposure, population density and nighttime-light intensity were integrated with the susceptibility results to construct a natural–social flood risk framework. This observation-based and explainable approach demonstrates the applicability of multi-source remote sensing for flood assessment in data-scarce regions, offering a robust scientific basis for flood management and spatial planning in monsoon-affected areas. Full article