Search Results (12,203)

Point-of-interest (POI) recommendation in location-based social networks (LBSNs) predicts a user’s next visit using check-in records. As critical factors influencing user decision making, geographic and social contexts are incorporated to deliver higher-quality recommendation services. However, current methods suffer from two limitations: geographical modeling is confined to distance thresholds, ignoring long-range spatio-temporal transitions, and social graphs remain static, failing to capture dynamic behavioral similarities among unconnected users. To address these gaps, we propose GSTRDFA (Global Spatio-Temporal Relationship- and Dynamic Friendship-Aware), a model comprising three layers. First, we construct spatio-temporal KGs (STKGs) that encode four relationship types: global spatio-temporal and local geospatial POI–POI links, dynamic user–user friendships, and static social ties. Second, four dedicated encoders—STSEncoder (spatio-temporal state embedding), GeoEncoder (geographical convolution), DFEncoder (graph attention network), and SocEncoder (GraphSAGE)—propagate and aggregate user and POI embeddings along these STKG relations. Third, a GRU-based sequence predictor uses the fused embeddings to match candidate POIs to the user. Evaluations on Foursquare datasets (NYC, JK, CA) show that GSTRDFA outperforms existing methods, improving Acc@1/5/10 and MRR by 0.24–3.31%. Key contributions include (1) unifying spatial, temporal, and dynamic social signals via STKGs; (2) jointly modeling global spatio-temporal transitions and dynamic friendships; and (3) enabling balanced short-/long-range and short-/long-term transition prediction. Full article

3D Gaussian splatting (3DGS) has recently emerged as an efficient and scalable method for high-fidelity 3D scene reconstruction, representation, and real-time rendering. In addition to geometric reconstruction, increasing research attention focuses on enriching 3D Gaussian primitives with semantic information, which can be related to an arbitrary application or domain, as well as common knowledge. However, the existing approaches to semantic 3DGS significantly differ in how semantics are represented, learned, and accessed, which makes systematic analysis difficult. This paper provides a review on semantic extensions to 3DGS. We introduce a unified multi-axis taxonomy that enables us to classify the available methods in terms of five complementary categories: semantic vocabulary space, representation form, functional role, knowledge source, and query mechanism. The analysis reveals key design trade-offs related to the flexibility, efficiency, and semantic expressiveness of the methods. Furthermore, we review datasets, benchmarks, and evaluation metrics used in the field, indicating the diversity of approaches and the lack of common evaluation frameworks. Based on this analysis, we also identify open challenges and possible future research directions. The presented survey is relevant to advances in games and immersive technologies, where semantically enriched real-time 3D representations are essential for interactive environments, AR/VR, and intelligent scene understanding. The systematic analysis presented in this survey aims to facilitate a deeper understanding of semantic 3DGS and support the development of more general, efficient, and task-aware 3D scene understanding systems. Full article

When a system must detect several related conditions over the same inputs, its designer can train a specialized classifier for each condition or one pooled model for all of them. Published comparisons of these designs contradict one another, and certain dataset properties, class imbalance above all, usually take the blame. We re-examine the comparison on code-smell detection, a domain where the contradiction is well documented, testing three ways of building the pooled baseline across five code-smell datasets, four generic benchmarks, four classical classifier families, and a neural network, with matched cross-validation folds and threshold-independent scoring. Much of the disagreement turns out to be manufactured by the comparison itself. The conventional pooled baseline appends a task indicator to the features, which quietly forces linear models, and neural networks conditioned only through the input, to share one set of feature weights across tasks; tree ensembles and softmax classifiers, which carry per-task structure of their own, are unaffected. Fixed-threshold F1 scoring accounts for the rest. Once the pooled model receives per-task weights, the gap closes on every dataset, and controlled positive-rate sweeps show that class imbalance moderates nothing. For tabular task families of this kind, the choice between the designs is an engineering decision, not an accuracy one; we close with a short protocol for making such comparisons fairly. Full article

Thermal drying is a critical phase in coffee processing, significantly influencing energy consumption, moisture uniformity, storage stability, and sensory quality. Despite the superior throughput of mechanical dryers over open-sun drying, these systems often operate with limited observability and manual control. This systematic review evaluates the readiness of mechanical coffee drying for Digital Twin (DT) integration. A comprehensive search across Scopus, Web of Science, IEEE Xplore, and ScienceDirect identified 22,859 records. Following multi-stage screening, 58 studies were retained for qualitative synthesis, categorized into a primary coffee-drying corpus and a secondary transferable corpus of methods from related food-processing applications. Findings indicate that while DT-enabling components, such as CFD models, drying-kinetics models, IoT monitoring, and non-destructive sensing, are established, they remain fragmented. No fully implemented and operationally validated DT for mechanical coffee drying was identified. Based on the evidence, a hybrid reduced-order physics-based model integrated with constrained supervisory control represents the most defensible near-term architecture. Future research should prioritize standardized datasets, uncertainty-aware soft sensors, and field validation across diverse dryer topologies and operating conditions. Full article

Pancreatic ductal adenocarcinoma (PDAC) is associated with high mortality owing to late diagnosis, aggressive tumor biology, and limited clinically useful early diagnostic and prognostic biomarkers. This study provides an exploratory evaluation of the expression profile, survival associations, co-expressed genes, functional association networks, and regulatory relationships of MEAK7 (KIAA1609/TLDC1) in pancreatic cancer using multi-platform bioinformatic approaches. We conducted a comprehensive multi-platform bioinformatic analysis using multiple publicly available datasets and platforms, including TCGA/GTEx, GEO, GEPIA3, HPA, CPTAC/UALCAN, TNMplot, Kaplan–Meier Plotter, DoSurvive, TIMER 3.0, STRING, TargetScan, miRDB, ENCORI and lncRNADisease. Gene expression patterns, survival associations, immune cell infiltration and regulatory non-coding RNA networks were systematically investigated. A STRING-derived protein functional association network was evaluated. MEAK7 expression patterns were additionally assessed in external GEO datasets, including GSE62165, GSE71729, and GSE183795. In summary, MEAK7 expression was elevated in pancreatic cancer compared with normal pancreatic tissues, was detectably expressed across multiple pancreatic cancer cell lines, and was increased in patient-derived PDAC samples. ROC analysis showed that MEAK7 expression discriminated PDAC tumor tissues from adjacent non-tumor pancreatic tissues, with an AUC of 0.795 (95% CI: 0.719–0.871) indicating exploratory transcriptomic discrimination rather than clinical diagnostic performance. Exploratory survival analyses showed that elevated MEAK7 expression was associated with unfavorable overall survival and disease-free outcomes, while multivariable Cox analyses showed that this association persisted after adjustment for the covariates available within the analyzed dataset. GEO-based analyses supported increased MEAK7 expression in early-stage pancreatic cancer samples, although metastatic expression patterns varied across datasets. Correlation and interaction analyses revealed positive associations between MEAK7 and PDAC-related genes, including CTTN, CDCP1, SMARCA4, SFN, and PALB2, as well as high-confidence STRING functional associations with V-ATPase components and RNASEK. Furthermore, MEAK7 was positively correlated with lncRNA UCA1 and negatively correlated with miR-582-5p. Collectively, these findings identify reproducible expression and survival-associated patterns involving MEAK7 across the analyzed datasets and provide a hypothesis-generating framework for further investigation of its potential biological role in PDAC. Experimental studies and prospectively characterized patient cohorts are required before any clinical, prognostic, diagnostic, or therapeutic implications can be established. Full article

Background: Medication non-adherence is a major driver of poor therapeutic outcomes, and computer vision methods that observe facial movements offer a non-contact route to verifying oral medication intake. Objective: To map and synthesize the existing literature on computer vision techniques applicable to the monitoring of medication intake, focusing on face tracking methods, oral movement detection and deglutition recognition, and to assess their potential in supporting automatic medication adherence verification systems. Eligibility criteria: Peer-reviewed articles, conference papers, patents, theses and preprints published from 2016 onward, written in English or Portuguese, applying facial landmark tracking or face analysis to ingestion-related movements (mouth opening, hand-to-mouth motion, pill placement, mastication or deglutition); studies confined to object/pill detection without facial analysis, or to general food intake without transferability to medication, were excluded. Sources of evidence: A systematic screening of 362 initial records was conducted across six main electronic databases and repositories: Google Scholar, PubMed, ScienceDirect, arXiv, IEEE Xplore, and Espacenet. Charting methods: Data were charted with a standardized, pilot-tested extraction form capturing bibliographic attributes, dataset type, experimental environment, face tracking approach, tools/models, and target movements; extraction was performed by a single reviewer. Following the screening process, a final selection of 34 relevant studies was included for detailed analysis and mapping. Results: Among the 34 included studies, 14 employ facial landmarks, 11 utilize temporal deep learning models, 6 apply facial action models and 3 rely on hybrid multimodal approaches that combine video analysis, object detection and temporal modeling. Tasks such as detecting mouth opening or tracking pill-to-mouth movement show promising results, while accurately detecting deglutition remains a technical challenge due to high sensitivity and individual variability. Limitations: The majority of the literature relies on private or institutional datasets (31 studies) and operates in controlled laboratory environments (22 studies); only 2 studies evaluated their methods via independent external datasets, which limits the generalization of current solutions to real-world telemonitoring scenarios. Conclusions: The literature indicates the existence of solid technical foundations for developing automated medication intake verification systems. To advance the field toward practical deployment, future research must address the need for more diverse datasets, real-world validation and more robust, adaptable modeling frameworks. Full article

Agricultural technological innovation is essential for sustainable agricultural modernization and rural development. However, agricultural technology enterprises often face persistent financing constraints because research and development (R&D) activities involve long investment cycles, high uncertainty, and delayed returns. Using a firm-level panel dataset of Chinese agricultural technology enterprises, this paper examines the effect of relational patient capital (RPC) on agricultural technological innovation by employing a two-way fixed-effects model. The results show that RPC significantly promotes agricultural innovation output. Mechanism analysis indicates that RPC enhances innovation through two channels. First, it facilitates firms’ digital transformation, thereby reducing R&D uncertainty and organizational costs. Second, it alleviates financing constraints by stabilizing cash flows to support R&D investment. The results remain robust after clustering standard errors, excluding the years affected by the COVID-19 pandemic, and employing lagged specifications. Heterogeneity analyses further reveal that the positive effect is more pronounced among small-scale enterprises and firms located in central and eastern China, where financing frictions and resource constraints are relatively more severe. By linking RPC to firm-level agricultural innovation, this study extends the literature on agricultural finance and innovation financing, highlighting the role of long-term, relationship-based capital in addressing market failures in agricultural R&D. The findings suggest that rural financial policies should encourage stable, long-term investment, strengthen financing support for small agricultural technology enterprises, and integrate patient capital with digital transformation initiatives to promote sustainable agricultural and rural development. Full article

Photoplethysmography (PPG)-based blood glucose estimation is attractive for non-invasive monitoring, but its development is constrained by limited paired PPG–glucose data and the weak representation of glucose-related waveform variations. This study proposes a framework combining conditional diffusion-based data augmentation with a cascaded multi-view attention bidirectional long short-term memory (BiLSTM) network. Blood glucose level, heart rate, and body mass index were used as physiological conditions to generate synthetic PPG segments, while channel and temporal attention together with cascaded BiLSTM layers were used to extract discriminative spatiotemporal features. A moth–flame optimization algorithm was employed to tune key hyperparameters. Experiments were conducted on a public dataset containing 67 PPG recordings from 23 participants. The framework was evaluated using subject-wise five-fold cross-validation, ensuring that all recordings from the same participant remained within a single fold. In the primary seed-42 analysis, the proposed method achieved a recording-level RMSE of 0.59 mmol/L, an MAE of 0.38 mmol/L, a MARD of 5.48%, and Clarke Zone A and A + B proportions of 97.0% and 100%, respectively; repeating the complete evaluation with seeds 2026 and 3407 produced closely similar recording-level results. These results suggest that physiologically conditioned generative augmentation may improve PPG-based blood glucose estimation in small-sample settings. Full article

Background: Attention-Deficit/Hyperactivity Disorder (ADHD) affects an estimated 6% of adults in the United States and contributes to a significant economic burden. Telehealth has emerged as a vital tool in the management of ADHD, offering improved access to care, especially for individuals in underserved communities. Despite its growing role, there remain critical gaps in understanding how social determinants of health (SDOH) are associated with disparities in telehealth utilization for ADHD treatment. Objectives and Methods: This study analyzed data from the National Center for Health Statistics (NCHS) Rapid Surveys System (RSS) Round 2: ADHD (October–November 2023), a nationally fielded survey of U.S. adults. Respondents were classified into three groups: never diagnosed, previously diagnosed, and currently diagnosed with ADHD. The study aimed to (1) compare the distribution of SDOH across ADHD status groups and the general adult population to identify factors associated with ADHD diagnosis; (2) assess the homogeneity of SDOH distributions across ADHD groups; (3) evaluate telehealth utilization among adults currently diagnosed with ADHD; and (4) examine the relationship between SDOH and telehealth use for ADHD treatment. Multivariable logistic regression (MVLR) served as a benchmark model, while machine learning (ML) models—including regularized linear regression, support vector machine (SVM), random forest (RF), LightGBM, multilayer perceptron (MLP), and Few-Shot Learning (FSL)—were trained to identify key predictors. Results: A total of 7009 survey responses were analyzed: 124 had a past diagnosis, 444 were currently diagnosed, and the remainder had never been diagnosed with ADHD, corresponding to a current ADHD prevalence of 6.3%. Adults with current ADHD were more likely to be male, single, younger, white, non-homeowners, and frequent users of online health resources. They also reported lower education, income, and financial security. About 70% used telehealth for counseling and prescriptions; insurance covered telehealth visits for 82.32% of users, yet 38.76% reported no coverage of ADHD-related diagnostic or treatment costs. Nineteen SDOH elements across four domains—demographic, socioeconomic, neighborhood/built environment, and healthcare access—were identified as predictors. ML models outperformed MVLR, with SVM and FSL achieving the highest F1 (both 0.63), and FSL the highest recall (0.69). Age, race, marital status, difficulty paying bills, home ownership, education, and household size were the most consistently important variables. Limitations: This study is limited by a cross-sectional design, reliance on self-reported ADHD diagnoses, and a lack of genetic or family-history measures. Additionally, the omission of complex sampling weights limits the national representativeness of these findings. Finally, the small effective sample size poses risks of model overfitting, and the generalizability of the models could not be externally validated due to the unavailability of comparable independent datasets. Conclusions: Despite widespread internet access, disparities in telehealth use for ADHD persist. Among 19 SDOH predictors, age (aOR = 0.56), difficulty paying medical bills (aOR = 2.52), and race (aOR = 1.37) were significantly associated with telehealth use, and all ML models outperformed the MVLR benchmark, though bootstrap CIs overlapped. Future research should incorporate inclusive data collection and stratified modeling to better represent disadvantaged populations and inform equitable access strategies. Full article

Young adults are often treated as a single digital consumer segment, although adjacent age cohorts may differ in e-commerce use and attitudes toward circular consumption. This study compares respondents aged 18–24 and 25–34, and examines whether age cohort or sustainability orientation is more closely associated with e-commerce behaviors and circular product acceptance. The analysis uses a respondent-level dataset harmonized ex post from five separate convenience-sample questionnaire surveys (n = 482 young adults). Respondents were not linked across sources; each hypothesis was tested on the subsample with the relevant variables observed, and the sustainability and circular-acceptance tests draw on n = 179 respondents from two of the five sources. Analyses used Mann–Whitney U tests, Spearman correlations, Benjamini–Hochberg false discovery rate correction, robust ordinary least squares models, and selected ordinal logistic models. Respondents aged 25–34 reported more frequent online shopping, whereas respondents aged 18–24 showed more positive personalization attitudes only before controls were introduced. No meaningful age differences were found in sustainability orientation, ecological purchase willingness, or willingness to buy refurbished products, while sustainability orientation was consistently associated with both circular-acceptance outcomes. The Age-Differentiated Circular Acceptance in Retail E-commerce framework is proposed as a conceptual synthesis: age relates to selected digital routines, while sustainability orientation is a cross-cohort correlate of circular product acceptance. The findings suggest that retailers should prioritize sustainability orientation, product-condition transparency, and risk-reducing information over broad age-based segmentation. Full article

This study examines the impact of energy security risk on the economic growth rate of the world’s 74 largest energy-consuming countries. The Energy Security Risk Index is employed to assess national vulnerability to fluctuations in energy security risk over a specified period. To investigate the empirical relationship between energy security risk and economic growth, the analysis utilizes a balanced panel dataset spanning from 1980 to 2025 and applies a series of econometric techniques, including cross-section dependence tests, unit root tests, Pedroni’s residual cointegration test, pooled mean group estimation, and a heterogeneous panel causality test. The empirical findings confirm the presence of a statistically significant causal relationship running from energy security risk to economic growth (LESR → GDPG) at the aggregate panel level. In advanced economies, energy security risk is found to constrain economic growth. Conversely, in developed economies, the results indicate that energy security risk exerts a positive and statistically significant effect on economic growth, which may be attributable to energy-related investments that drive structural transformation. These findings carry important implications for the global community, given that energy continues to serve as the lifeblood of modern production systems. Full article

Accurate carbon footprint accounting is fundamental for urban environmental governance. However, multi-stakeholder transit networks struggle with data manipulation, privacy risks, and labor-intensive manual auditing. To resolve these trust and scalability bottlenecks, this paper introduces a tri-layer hybrid blockchain framework based on an “off-chain storage, on-chain evidence” paradigm. The architecture synergizes a relational database (MySQL) for high-throughput structured data, the InterPlanetary File System (IPFS) for decentralized raw evidence, and Hyperledger Fabric to immutably anchor dual-layer cryptographic hashes. We engineer a smart contract auditing pipeline that autonomously executes deterministic verification of hash consistency, emission thresholds, and physical logic integrity. Empirical evaluations utilizing a large-scale urban transit dataset injected with adversarial mutations demonstrate high robustness, achieving F1-scores of 1.000 across multidimensional anomalies. This replaces manual testing with statistically significant verification. Ultimately, this framework provides environmental regulators and transit authorities with a highly scalable, privacy-preserving, and trust-minimized infrastructure for continuous carbon footprint traceability. Full article

Dementia-related disorders, particularly Alzheimer’s disease (AD), represent a growing global health challenge, increasing the need for early and reliable detection. Electroencephalography (EEG), a non-invasive and cost-effective neurophysiological modality, has emerged as a promising tool for identifying neural signatures associated with cognitive decline. Recent advances in machine learning (ML) and deep learning (DL) have enabled more effective analysis of complex EEG signals for automated dementia prediction. This survey provides a comprehensive synthesis of EEG-based dementia studies published between 2020 and 2025, with a primary focus on Alzheimer’s disease (AD), frontotemporal dementia (FTD), mild cognitive impairment (MCI), and related dementia disorders. Unlike previous reviews that emphasize multimodal approaches or specific methodologies, this work exclusively focuses on EEG and presents a systematic comparison of ML and DL approaches, preprocessing pipelines, feature extraction techniques, publicly available datasets, validation strategies, performance metrics, and explainable methods. We further examine current research trends, identify methodological limitations such as small dataset sizes, subject-level data leakage, class imbalance, inconsistent preprocessing protocols, and limited multicenter validation, and discuss their implications for model generalizability and clinical adoption. We conclude by outlining future research directions toward developing robust, interpretable, and scalable EEG-based dementia prediction systems suitable for real-world clinical applications. Full article

RNA large language models (RNA LLMs) show potential for predicting viral RNA–RNA interactions (RRIs), but their ability to transfer between viruses remains unclear. Here, we evaluated three nucleotide language models using vRIC-seq-derived RRI datasets from two members of the Coronaviridae family, SARS-CoV-2 and porcine deltacoronavirus (PDCoV). We compared within-virus five-fold cross-validation with bidirectional cross-virus testing. DNABERT achieved within-virus AUC values of 0.9595 for PDCoV and 0.9741 for SARS-CoV-2. Under cross-virus transfer, its AUC decreased to 0.8372 when trained on PDCoV and tested on SARS-CoV-2 and to 0.8116 in the reverse direction, with corresponding F1 scores of 0.7847 and 0.7617. RNAErnie retained similar but slightly lower cross-virus discrimination, whereas the Nucleotide Transformer approached random discrimination. These results provide a two-virus proof of concept that sequence-based models can retain partially shared discriminative signals across coronavirus genera. They do not establish functional, tropism-related, or receptor-dependent transferability, and broader evaluation across additional viruses and orthogonal structural validation will be required before application to emerging-virus screening. Full article

Detecting small anomalous targets in hyperspectral imagery is challenging when neither target spectra nor spatial morphologies are known, because most unsupervised detectors still encode implicit assumptions about background statistics or target structures and therefore generalize poorly to genuinely unknown threats. This paper introduces the Hard-Negative-Guided Topology Transfer Network (HNGT-Net), a lightweight teacher–student framework that addresses this problem from two perspectives. On the representation side, a similarity graph coupling spatial adjacency with feature-space nearest neighbors characterizes the spectral–spatial topology of normal backgrounds, and a three-level consistency objective over node embeddings, structural relations, and graph-Laplacian responses transfers this topology from a frozen teacher to a compact student. On the discrimination side, hard negatives are synthesized directly on pure normal pixels through sparsity-gated projected gradient perturbation, while a response-margin constraint forces the student to score such negatives above normal samples and mitigates distribution overfitting. Experiments on four public benchmark scenes and one Salinas-derived synthetic dataset show that HNGT-Net achieves an average AUC of 0.9953 with the smallest variance among all competitors. The student branch provides 10.5-fold parameter compression relative to the teacher, while the complete teacher–student stack remains compact, supporting resource-constrained remote sensing deployment. Full article