Search Results (12,807)

The early detection of Severe Combined Immunodeficiency (SCID) and X-linked agammaglobulinemia (XLA) prevents fatal outcomes. This study presents the first pilot TREC/KREC newborn screening (NBS) program in southern Kazakhstan, a high-birth-rate region, to establish local reference ranges and assess operational viability. A multiplex real-time PCR assay was used to quantify T-cell (TREC) and kappa-deleting recombination excision circles (KRECs) from dried blood spots of 5000 unselected neonates. Biomarkers were normalized to copies per ${10}^6$ cells using albumin as a diploid reference gene. Regional 0.5th percentile cut-offs were established (TREC < 3165 copies/${10}^6 \mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}\mathrm{s}$ and KREC < 2554 copies/${10}^{6 }$ cells), and gender and gestational age did not significantly impact biomarker levels. While a low birth weight (≤2500 g) significantly reduced KREC levels, the extreme lower distribution tails remained unaffected, validating the use of universal, unstratified thresholds. Applying these cut-offs yielded an optimal 1.0% initial recall rate. Consistent with global incidence rates, no true positive cases were identified. The established assay and universal percentile cut-offs demonstrate high levels of analytical reliability and demographic stability. This pilot confirms the regional pediatric healthcare infrastructure’s readiness for a routine, population-based NBS program without the need for complex algorithms. Full article

Ride-hailing global positioning system (GPS) mobile probe data provide flexible urban traffic observations, but their sparse and uneven coverage makes model evaluation difficult because observed targets, valid predictions, and historical input support do not always coincide. This study audits ultra-short-term road-segment speed forecasting under sparse mobile sensing using a mask-consistent support-chain framework. A three-day GPS dataset is aggregated into 5 min speed observations over 1970 road segments and used as a controlled sparse-sensing case study rather than a general-purpose long-term forecasting benchmark. The evaluation protocol distinguishes the full test grid, the set of directly observed target speeds, model-valid prediction support, strict complete-history support, and common-support subsets for coverage-limited baselines. The directly observed target set is used as the primary relaxed support because it retains all verifiable ground-truth targets, while strict and common-support subsets are reported as sensitivity checks. Under this support-conditioned evaluation, the adaptive graph convolutional recurrent network (AGCRN) is associated with lower mean absolute error (MAE) among full-coverage models, the historical mean (HIST_MEAN) baseline is associated with lower root mean squared error (RMSE), and congestion recall remains below 0.24 for all full-coverage deep models. These complementary results indicate conditional and metric-dependent strengths rather than universal model superiority. Because the dataset covers only three consecutive days, weekday/weekend variation, incident-specific fluctuations, seasonal effects, and spatial transferability cannot be fully examined and are treated as limitations. Overall, the findings show that evaluation support should be reported as a first-order experimental factor alongside model accuracy under sparse mobile probe sensing. Full article

Retrieval-Augmented Generation (RAG) grounds large language models in external evidence, but many RAG systems represent knowledge either as flat text chunks or as automatically constructed indexing graphs. This assumption is incomplete for curated thematic corpora, including religious scriptures, legal codes, clinical guidelines, educational taxonomies, policy documents, and library classification systems, where domain experts have already organized knowledge into thematic paths and citeable canonical units. This paper investigates how RAG can exploit such expert-authored structures while pruning evidence to a compact and query-specific set. We conduct a critical survey supported by a bibliometric analysis of 2815 Scopus-indexed RAG-related records exported on 26 May 2026, of which 2809 records were retained after duplicate removal. The bibliometric results indicate rapid growth in RAG research but limited explicit consolidation around curated thematic paths, canonical evidence units, or thematic path-guided evidence pruning. We therefore propose ThemePath-RAG, a retrieval framework that retrieves curated thematic paths as high-recall semantic routes, expands candidate canonical evidence, and applies query-aware scoring and global pruning before generation. To assess operational feasibility, we implement ThemePath-RAG for Qur’anic question answering and compare it with a Vector RAG baseline on 150 paired questions using RAGAS context relevance with gpt-4o-mini as the LLM evaluator. Both methods return approximately three final ayat per question. Vector RAG achieves higher mean context relevance than ThemePath-RAG ($0.920$ versus $0.798$; $p<0.001$). Thus, the proof of concept establishes the feasibility of thematic-path-guided retrieval and identifies evidence-selection challenges, rather than demonstrating superiority over conventional vector retrieval. The paper clarifies the framework’s relationship to GraphRAG, LightRAG, HippoRAG, PathRAG, ontology-based RAG, and AI-augmented bibliometric systems, and outlines a language-matched, multi-baseline evaluation agenda for future cross-domain validation. Full article

Traditional manual grading of fresh chili peppers suffers from inconsistent quality control and low efficiency. To meet the demand for accurate fruit shape recognition during the post-harvest stage, this study proposes an intelligent recognition method based on an improved DenseNet-121 network. This approach facilitates the application of machine vision in agricultural sorting equipment. DenseNet-121 serves as the backbone network. The Convolutional Block Attention Module (CBAM) is introduced to enhance feature focus on fruit shapes. A regularization strategy (Dropout = 0.3, weight decay = 1 × 10⁻⁴) and a cross-entropy loss function with label smoothing (LS = 0.1) are integrated to optimize decision boundaries. These configurations prevent the model from overfitting to hard training labels and yield a robust classification architecture. Experimental results demonstrate that the proposed model achieves a precision of 90.09%, a recall of 89.60%, an F1-score (the harmonic mean of precision and recall) of 89.53%, and an overall accuracy of 89.74%. The model contains 7.09 M parameters and requires a single-frame inference time of 7.35 ms. Comprehensive evaluations indicate that the proposed model achieves an optimal balance among environmental noise robustness, prediction accuracy, and computational efficiency. Consequently, by maintaining high fine-grained classification accuracy alongside a low memory footprint and rapid inference speed, the model demonstrates strong potential for real-time deployment on resource-constrained edge devices within actual agricultural optical sorting equipment. Full article

Hundreds of connected components expose vehicle systems to an increasing number of cyber attacks. Vehicle manufacturers must establish the most appropriate procedures for vulnerability assessment and penetration testing using a mix of proprietary solutions and open-source standards. The spread of Large Language Models (LLMs) simplifies the interaction between automotive experts and domain-specific knowledge bases. While proprietary LLM services can be expensive and raise data privacy concerns, open-source LLMs are potentially more cost-effective and better suited to in-house solutions. However, the effectiveness of open-source models in retrieving automotive-related cybersecurity information remains unclear. While adopting open-source LLMs with a few billion parameters, their reasoning and generative capabilities under in-context learning settings are questionable. To bridge this gap, this paper explores efficient solutions for Retrieval-Augmented Generation (RAG) architecture for automotive cybersecurity relying on open-source LLMs. The ultimate goal is to enable cost-effective retrieval and question answering from in-domain knowledge bases, overcoming the privacy and confidentiality issues raised by automotive experts. Using a Graph Knowledge Base designed for a corporate scenario, this paper first defines an expert-curated testing benchmark to evaluate in-domain question-answering performance across multiple aspects. Next, it proposes different RAG system variants based on various retrieval strategies and LLMs, both proprietary and open-source. Finally, it quantitatively evaluates the effectiveness of the content retrieval strategies and compares the pertinence, conciseness, and completeness of generated answers through human validation. Notably, within the scope of the performed analysis, RAGs that rely on open-source models demonstrate promising and competitive performance in some respects compared to the OpenAI GPT model. RAG retrieval performance also surpasses that of state-of-the-art solutions on existing cybersecurity benchmarks (Recall@K above 0.95 vs. 0.65 for state-of-the-art in-domain RAGs). Full article

Alzheimer’s Disease (AD) necessitates high-resolution transcriptomic biomarkers for early detection, yet current computational methods are hampered by high-dimensional search space and publication bias regarding imbalanced datasets. We propose the Hybrid Filter–Wrapper Nomadic People Optimizer, a three-stage pipeline integrating a tri-criterion filter, an enhanced NPO wrapper with adaptive Lévy-scale anti-stagnation mechanism, and a five-member soft-voting ensemble. The system was evaluated using a dual-paradigm protocol; Scenario A (balance brain tissue; GEO dataset GSE 33000, GSE 132903, GSE122063) and Scenario B (imbalanced peripheral blood: GSE 63060 + GSE 636061). In scenario A, HFW-NPO outperformed 13 published methods, achieving balanced accuracy of 85.28%, 87.16%, and 96.67% while identifying compact panels of 29–32 probes per fold (observed range: 24–38). Scenario B, evaluated on a merged 478-samples peripheral blood cohort (GSE63060 + GSE 636061 imbalanced 1.48:1) with z-score batch harmonization and RSKF (5 × 10) cross-validation, achieved a balanced accuracy of 59.53% and MCI Recall of 63.50 ± 14.02%, providing the first reproducible baseline for this clinically challenging task, while acknowledging that 59.53% balanced accuracy does not yet reach clinically actionable levels. By providing transparent reporting across both balanced and severely imbalanced datasets, this study establishes a state-of-the-art, reproducible framework for AD biomarker discovery and provides a critical baseline for the challenging task of transcriptomic-based classification in peripheral blood samples. Result is currently scoped to Illumina HumanHT-12 microarray data, and cross-platform validation on RNA-seq cohorts is identified as a priority future extension. Full article

Hoof slipping in dairy cows is a subtle, transient hoof motion event distinct from lameness or falling, with short duration, limited displacement, and close resemblance to normal gait, making automated detection particularly challenging; relevant methods remain scarce. This study proposes a cascaded detection framework based on improved DeepLabCut and NeuFlow v2 for automated hoof-slipping detection and distance estimation in Holstein dairy cows. The four-stage framework covers hoof key point localization, pixel-level optical flow fusion, motion parameter curve feature extraction, and Random Forest classification. The framework was developed on Dataset 1, which contained 115 single-cow side-view videos. Of these, 31 contained slipping events and 84 were normal walking. It was further assessed on a smaller second-farm dataset of 17 single-cow videos (Dataset 2). ResNet-50 with a Coordinate Attention mechanism was adopted as the backbone, reducing mean four-hoof localization RMSE to 2.80 pixels across five independent training runs, showing a 15.2% improvement over the baseline, and outperforming YOLOv8s-Pose. NeuFlow v2 was applied to extract the localized optical flow from hoof regions, yielding velocity and directional curves from which slipping features were derived. The Random Forest classifier achieved an accuracy of 98.9%, precision of 93.3%, recall of 90.3%, F1 score of 91.8%, and AUC of 0.995, outperforming MViT, SlowFast, and STME. The slipping distance estimation RMSE was 1.22 pixels. With the localisation model retrained on new farm frames, the method reached comparable performance on the second farm, suggesting preliminary cross-farm generalisability that warrants larger-scale validation. The proposed framework provides a non-invasive basis for early hoof-health monitoring and welfare-oriented farm management. Full article

The rapid expansion of online health information channels, particularly emerging artificial intelligence (AI) platforms, is transforming how individuals access and evaluate health information. Drawing on an extended Comprehensive Model of Information Seeking (CMIS), this research examined how different channel types (AI-based, short-video, and text-based) influence online health information-seeking behavior (OHISB) through a pilot validation (N = 258), a cross-sectional survey (Study 1; N = 300), and a between-subjects experiment (Study 2; N = 300). Study 1 tested an extended CMIS model incorporating channel type, source credibility, information credibility, and perceived usefulness, while Study 2 examined the causal effects of channel exposure. Structural equation modeling in Studies 1 and 2 consistently showed that source and information credibility predicted OHISB indirectly through perceived usefulness. AI channels showed no advantage in Study 1, whereas Study 2 found that participants perceived AI sources as more credible and useful, which indirectly predicted stronger intentions for SAMC and information seeking through the credibility–usefulness pathway. This change may reflect methodological differences between self-report recall-based and direct exposure designs, and the public’s growing familiarity with AI technologies. By integrating channel characteristics and credibility perceptions, this study extends the CMIS framework and provides evidence for AI’s enhanced perceived credibility in health information contexts, offering insights for improving AI-driven health communication. Full article

Background/Objectives: Accurate estimation of meal composition from food images can support safer and more reliable insulin bolus decision-making for individuals with Type 1 diabetes. Existing food recognition and nutrition estimation systems are often designed for general dietary logging and do not directly integrate food analysis with personalized insulin therapy parameters. Methods: This study presents an image-based nutrition estimation and insulin decision-support module developed within the AI-assisted Diabetes Care (AIDCARE) platform. The proposed system uses a convolutional neural network (CNN) to classify food items from a single meal image, and retrieves reference nutritional values from a food composition database. A separate multimodal large language model (MLLM)-based estimation component is then used to estimate portion size, allowing carbohydrate and nutrient values to be scaled according to the observed serving. Results: A curated food image dataset containing 40 food categories was used to evaluate three CNN architectures: ResNet50, Inception V3, and EfficientNet-B0. EfficientNet-B0 achieved the best classification performance, with 94.91% validation accuracy, 95.55% precision, 94.87% recall, and 94.90% F1-score. The portion-estimation component achieved an MAE of 12.27 g and an RMSE of 15.11 g. The estimated carbohydrate value is combined with user-specific clinical parameters, including the insulin-to-carbohydrate ratio and insulin sensitivity factor, to generate advisory bolus guidance. To support safety, the system requires user confirmation or correction of the recognized food category and estimated portion before insulin guidance is displayed. Conclusions: The proposed system is intended for advisory decision support only and is not designed to replace clinical judgment or autonomous insulin delivery systems. Full article

Grasslands represent one of the most extensive terrestrial biomes globally, covering approximately one-third of the Earth’s land surface, yet they are increasingly threatened by land-use change and overgrazing, underscoring the need for reliable monitoring approaches. This study compares the performance of machine learning and deep learning algorithms for grassland mapping using multi-source remote sensing data derived from Sentinel-1, Sentinel-2, and terrain variables. The research was conducted in Mpumalanga Province, South Africa, a heterogeneous landscape comprising lowland savannas, high-altitude grasslands, escarpments, and riverine wetlands. Random Forest (RF) and Support Vector Machine (SVM) classifiers were implemented in Google Earth Engine using fused satellite and terrain datasets with field-collected samples for training and validation, while a One-Dimensional Convolutional Neural Network (1D-CNN) was developed in Python 3.13.5 using the same inputs. Results demonstrate that integrating multi-source data improves classification accuracy, with radar-based features contributing the most. RF achieved the highest performance, with an overall accuracy of 97.7% and grass-class precision, recall, and F1-score exceeding 0.97, closely followed by the 1D-CNN with 91% overall accuracy and complete grass detection. In contrast, SVM performed notably lower with an overall accuracy of 80,8%. These findings highlight the effectiveness of advanced learning approaches for grassland mapping and support their application in ecological restoration and environmental management. Full article

Coal and gas outburst is a major mine disaster affected by complex coupled factors, bringing obstacles to disaster prevention. To address low accuracy and poor generalization of traditional single-algorithm prediction models, this paper constructs a two-layer Stacking ensemble learning identification model for outburst risk. RF, SVM and AdaBoost serve as base models, and WOA-LightGBM acts as the meta-model. Based on measured data of a Shanxi coal mine, Spearman correlation analysis and RF dimensionality reduction remove redundant features; Borderline-SMOTE balances imbalanced samples with few severe-risk data. Accuracy, macro-precision, recall and F1-score evaluate model performance after parameter optimization. Results show that the proposed Stacking model reaches 0.9770 accuracy, outperforming single machine learning models and other intelligent algorithms. It presents minor index fluctuations with strong stability and correctly identifies all eight practical engineering cases. Combining feature engineering and Stacking learning effectively captures nonlinear relations between influencing factors and risk levels. The model owns high precision and robustness, offering reliable technical support for coal and gas outburst prediction and control. Full article

In social sensing environments, misinformation and coordinated manipulation often manifest through implicit semantic signals, complex behavioral dynamics, and highly coupled propagation structures. These factors pose significant challenges to artificial intelligence-driven sensing systems regarding data quality, multimodal fusion, and robustness. To address these issues, this study proposes an artificial intelligence-driven multi-granularity sensing framework. This framework integrates heterogeneous sensing signals from post-level semantic perception, user-level behavioral sensing, and group-level structural sensing into a unified representation space. Hierarchical consistency constraints enable cross-granularity sensing collaboration. This mechanism enhances stability and discriminative capability under complex and noisy data conditions. Methodologically, the framework jointly incorporates semantic sensing via text encoding, temporal sensing via behavioral sequence modeling, and structural sensing via graph neural network-based propagation. This integration effectively mitigates information bias induced by single-perspective sensing and improves the modeling of latent risk patterns. Experimental results on real-world datasets demonstrate that the proposed framework achieves significant improvements across multiple evaluation metrics. Specifically, it achieves a Precision of 0.847, a Recall of 0.812, an F1-score of 0.829, an Accuracy of 0.856, and an Area Under Curve of 0.913. It consistently outperforms traditional machine learning models, as well as mainstream deep learning and graph-based approaches. Furthermore, comparison experiments validate the complementarity among semantic, behavioral, and structural sensing signals. The full model achieves an improvement of more than 3 percentage points in the F1-score compared to single-granularity configurations. An ablation study further demonstrates that each sensing module contributes substantially to performance enhancement, with the semantic sensing and hierarchical consistency constraints playing particularly critical roles. Overall, the proposed method exhibits a strong capability to handle complex heterogeneous sensing data. It improves robustness and enhances cross-level information utilization, providing an effective solution to data-related challenges in artificial intelligence-driven sensing systems. Full article

Fall-from-height (FFH) detection is a critical component in wearable safety systems, particularly in environments where high-intensity movements can lead to frequent false positives. Conventional approaches based on simple thresholding of acceleration signals often fail to reliably distinguish FFH events from non-fall activities due to overlapping signal characteristics. This paper proposes a data-driven FFH detection method that integrates multiple complementary features into a unified score-based model. The proposed approach first performs structured peak detection to extract candidate impact events while significantly reducing the number of samples requiring further processing. Each candidate is then evaluated using pre-peak structure, post-impact stability, and pressure variation, which respectively capture structural, temporal, and physical characteristics of FFH events. Based on statistical analysis, feature-wise score contributions are designed to reflect their discriminative strength, and the final FFH decision is performed using an additive scoring mechanism. This formulation enables flexible handling of ambiguous cases while preserving strong FFH characteristics. Experimental results demonstrate that the proposed method maintains 100% recall at the selected decision threshold while significantly reducing false positives from non-FFH activities. In addition, the peak detection stage reduces more than 99% of raw samples, enabling efficient on-device processing suitable for wearable systems. The proposed method also includes quantitative analysis of latency characteristics. Although FFH inference latency is influenced by asynchronous pressure sensing, the delay remains bounded and predictable, and most detections are completed within a practical time range for real-time wearable safety applications. Overall, the proposed method achieves a practical balance between detection sensitivity, false-positive suppression, computational efficiency, and real-time feasibility, demonstrating its applicability to wearable safety systems. Full article

The continuous collection of road information by autonomous vehicles (AVs) has intensified regulatory pressure on data security protection. At present, the Chinese government adopts a proactive stance on protecting AV data security. Nevertheless, relevant requirements are scattered across various regulatory regimes, including data security, cybersecurity, personal information protection and AV access regulation. It has given rise to ambiguous judgement criteria and inconsistent law enforcement practices among local authorities. As a leading developed economy worldwide, the European Union has continuously refined legislation on data security protection for AVs since 2018, establishing a globally sophisticated protective framework. Against this backdrop, this paper adopts comparative analysis and normative analysis to focus on examining the characteristics of the EU’s advanced rules. The research reveals that the EU has integrated risk prevention, monitoring and reporting mechanisms into a unified regulatory framework. It implements market access and safety assessment before operation, conducts ongoing safety management during operation, and launches data reporting and recall procedures after operation. After evaluating the applicability of the EU model in China, this paper suggests China adopt phased AV data security rules. Governance should focus on early detection of data breach risks before operation, real-time data monitoring during operation, and data reporting after operation. This proposal clarifies responsibilities among regulators, automakers and data service providers, improves the predictability and enforceability of relevant governance and facilitates safe, innovative and sound development of the autonomous driving industry. Full article

Software defect prediction (SDP) aims to identify defect-prone modules before testing, but severe class imbalance and redundant software metrics often limit prediction performance. Many conventional feature selection methods estimate feature relevance with the original imbalanced empirical distribution and mainly emphasize marginal relevance or global classifier-oriented criteria, which may under-prioritize features that are informative for the minority defective class. To address this issue, this paper proposes SCAR-CMB, a simplified class-reweighted and interaction-aware feature selection method for imbalanced SDP. SCAR-CMB estimates feature-label dependency with a class-balanced empirical distribution, controls redundancy using weighted conditional dependency information, and incorporates an interaction-aware conditional-gain term as an auxiliary re-prioritization signal within a relevance-screened feature pool. Rather than performing full causal structure discovery or formal synergy estimation, SCAR-CMB adopts a Markov-blanket-inspired conditional dependency design as a practical guide for feature selection. The final configuration excludes both hardness-aware weighting and false discovery rate filtering. SCAR-CMB is evaluated on ten public NASA and PROMISE defect datasets under a leakage-free cross-validation protocol. Compared with seven representative baselines, SCAR-CMB achieves competitive overall performance and obtains the highest average defective-class recall, G-mean, and balanced accuracy. However, it is not uniformly superior across all metrics, and the recall advantage is not confirmed by the omnibus Friedman test. Additional mechanism-level, stability, and sensitivity analyses show that class reweighting changes feature prioritization, the selected feature subsets are relatively stable across folds, and the interaction-aware term provides limited and dataset-dependent auxiliary effects. Sensitivity analyses further indicate that the main conclusions are not solely determined by a specific feature budget, discretization-bin setting, or downstream classifier. Overall, SCAR-CMB should be interpreted as a practical minority-class-oriented feature selection method that provides a trade-off among defective-class detection, feature subset control, and computational cost. Full article