Guangdong Police College

Omega-3 polyunsaturated fatty acids (PUFAs) are known to modulate inflammatory signaling and enhance muscle recovery from exercise-induced stress. In this work, a meta-analysis was conducted that adhered to the PRISMA 2020 criteria, incorporating evidence from 41 randomized controlled trials on eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) supplementation, all of which had been conducted as evidence between 2011 and 2025 and evaluated the effects of EPA and DHA supplementation on the primary markers of the impact of exercise on inflammation and muscle injury, including the random-effects model with moderator dose, intervention duration, sex, and training status. The integrated findings showed that interleukin-6, tumor necrosis factor-α, creatine kinase, and delayed-onset muscle soreness (standardized mean difference = -0.4 to -0.7) were significantly and moderately reduced, indicating a uniform response to omega-3 supplementation's anti-inflammatory and recovery-promoting effects. C-reactive protein responses were more dispersed, suggesting that baseline inflammation differed and that the sampling protocols did as well. Subgroup analyses showed that doses of 2 g/day or more of mixed EPA + DHA, with a minimum duration of 6 weeks of administration, produced the strongest effects, especially among recreational athletes rather than elite athletes. Mechanistically, nuclear factor-kappa B activation and the ensuing synthesis of specialized pro-resolving mediators (resolvins, protectins, and maresins), as well as an increase in cellular antioxidant capacity, appear to be moderated by omega-3 supplementation and support efficient resolution of inflammation and tissue repair. These results form a body of evidence over more than 10 years, showing that omega-3 fatty acids are a strong, evidence-based nutritional tool for reducing the effects of post-exercise inflammation, supporting functional recovery, and sustaining long-term athletic performance.

This study aims to explore the interplay between parent-child communication, social support, physical exercise and non-cognitive abilities among adolescents. Using data from the China Education Panel Survey (2014-2015) with a nationally representative sample of 5,055 eighth-grade students. Analyses were conducted via Stata 17.0 and SPSS 26, employing chain mediation tests and bootstrap analysis. The results were as follows:1) parent-child communication directly contributes to the enhancement of adolescents' non-cognitive abilities, facilitates their access to social support, and promotes their participation in physical exercise; 2) parent-child communication indirectly improves adolescents' non-cognitive abilities through the chain mediating effect of social support and physical exercise; 3) the positive effect of parent-child communication on non-cognitive abilities is stronger among urban adolescents compared to rural adolescents, and more pronounced among only children than non-only children. In conclusion, parent-child communication not only directly and positively predicts adolescents' non-cognitive abilities but also exerts indirect effects via two single mediators (social support and physical exercise) and a chain mediation mechanism combining both mediators.

Although existing diffusion Transformer-based video generation models can synthesize high-quality dynamic portraits, they still face significant challenges in maintaining identity consistency across long sequences. We identify that the root cause lies in the model’s difficulty in decoupling stable identity features (such as facial structure) from transient appearance details (such as lighting and expressions) in reference images. To address this issue, we propose a lightweight Low-Rank Adaptation (LoRA) module based on Facial Structure Injection, which can be seamlessly integrated into pre-trained video generation models in a plug-and-play manner. Specifically, we first construct a portrait video-text dataset containing one million samples. Subsequently, we extract and render a "structure map" from the reference image that contains only the core facial topology, thereby eliminating the interference of appearance information. Finally, this structure map is utilized to guide the LoRA module, providing explicit and stable geometric priors to the base model at each step of the denoising process. Without any modifications to the large foundation model, our approach achieves significant improvements in identity preservation metrics by training only our compact LoRA module. Extensive experiments demonstrate that our method enhances identity consistency while maintaining the original model’s robust video fidelity and text controllability, providing a practical, efficient, and scalable solution for addressing identity drift issues in controllable portrait generation.

The Metaverse has been occupying people's attention in recent times, and relevant cases have appeared in many fields. It integrates related technologies such as the Internet of Things (IoT), blockchain, and artificial intelligence, and is developed on the basis of digital twins. However, the rapid increase in the number of IoT services and the security vulnerabilities of IoT devices have made the entire Internet of Things network and Metaverse face huge security threats. This paper designs a new shared intrusion detection framework for the IoT. It is based on the concept of intelligent sharing, enabling users in different locations to safely share their detection intelligence. Moreover, a universal intelligence description specification has been designed, which enables users to quickly and conveniently use the intelligence of others and easily realize collective learning. The simulation results verify that the proposed framework can improve the user's detection ability, realize collaborative learning, and overcome problems such as a lack of data.

This paper presents the design and evaluation of a real-time IoT-based emergency response and public safety alert system tailored for rapid detection, classification, and dissemination of alerts during critical incidents. The proposed architecture combines a distributed network of heterogeneous sensors (e.g., gas, flame, vibration, and biometric), edge computing nodes (Raspberry Pi, ESP32), and cloud platforms (AWS IoT, Firebase) to ensure low-latency and high-availability operations. Communication is facilitated using secure MQTT over TLS, with fallback to LoRa for rural or low-connectivity environments. A prototype was implemented and tested across four emergency scenarios fire, traffic accident, gas leak, and medical distress within a smart city simulation testbed. The system achieved such as consistent alert latency under 450 ms, detection accuracy exceeding 95%, and scalability supporting over 12,000 concurrent devices. A comprehensive comparison against seven state-of-the-art systems confirmed superior performance in latency, reliability (99.1% alert success), and uptime (99.8%). These results underscore the system’s potential for deployment in urban, industrial, and infrastructure-vulnerable environments, with future work aimed at incorporating AI-driven prediction and federated learning for cloudless operation.

In this paper, we investigate global strong solution and diffusive limit of the two-species Vlasov–Poisson–Boltzmann system in R3 for the full range of cutoff hard potentials 0 ≤ γ ≤ 1. By introducing a new Hx,v2−Wx,v1,∞ framework, which is based on the interplay between the velocity weighted Hx,v2 energy estimate and the time-velocity weighted Wx,v1,∞-estimate for the two-species Vlasov–Poisson–Boltzmann system, we construct the global weighted estimate uniformly in the Knudsen number ɛ ∈ (0, 1]. As a result, global strong solution is constructed and the two-fluid incompressible Navier–Stokes–Fourier–Poisson limit is rigorously justified for all cutoff hard potentials 0 ≤ γ ≤ 1.

Interleukin-6 (IL-6), a pleiotropic cytokine, is induced by infection of influenza A virus (IAV), where it plays a pivotal role in immune defense and the regulation of inflammation. IAV induces IL-6 transcription upon recognition by pattern recognition receptors, which activate downstream signaling cascades, leading to the activation of transcription factors. Activated transcription factors subsequently regulate the expression of IL-6 in innate immune cells, such as macrophages, dendritic cells, and epithelial cells. IL-6 contributes to antiviral immunity by promoting the recruitment of immune cells to sites of infection and amplifying the inflammatory response. While optimal IL-6 production is essential for effective anti-IAV immunity, excessive IL-6 production can contribute to a dysregulated immune response, leading to a cytokine storm. In this context, IL-6 signaling, in coordination with other proinflammatory cytokines such as TNF-α and IL-1β, not only enhances its own production but can also serve as a key mediator of inflammation. This cascade can result in exaggerated immune responses, causing tissue damage and potentially leading to severe outcomes, including organ failure and death. Understanding the molecular mechanisms underlying cytokine storms presents important therapeutic opportunities. However, the precise pathways responsible for excessive IL-6 production and its dysregulation during IAV infection is not fully understood. This review explores the reported mechanisms regulating IL-6 induction in response to IAV and its innate immune role in IAV pathogenesis, highlighting existing research gaps in understanding IAV-induced IL-6 production and its impact on immune modulation. A deeper understanding of IAV-induced IL-6 production and signaling could inform the development of targeted therapies to more effectively manage influenza.

ABSTRACTThe research on nested named entity recognition (NER) is conducive to providing richer semantic representations and capturing the nested structure among entities, which is crucial for the execution of downstream tasks. This paper aims to summarise the nested NER methods that have been combined with emerging technologies in recent years. We summarise the nested NER methods that are integrated with emerging technologies from three dimensions: model, framework, and data. Additionally, we explore the research progress of nested NER in two scenarios: cross‐lingual modality and multi‐modal in different modalities. Furthermore, we discuss the practical applications of NER technology in five fields: biomedicine, justice, finance, media, and e‐commerce. Through this review, we can clearly see the development trends of nested NER technology under emerging technologies and different modalities, as well as its broad application prospects in various fields. This provides a reference for future exploration directions in nested NER.

Lip print, an emerging biometric marker, can be used to identify suspects or victims in criminal investigations and forensic identification. However, problems such as difficulty in feature extraction, slow recognition speed, and low accuracy of lip print images remain. In this paper, we use the latest YOLOv10 image recognition model to identify and classify lip print images, and construct a lip print image dataset based on the CFD face dataset, evaluate the speed and accuracy performance of the differe nt models, and explore and study the deep learning models applicable to lip print recognition. The detection effect of the YOLOv10 model on the lip print image dataset is better than traditional models such as SSD, Faster R-CNN, YOLOv5, YOLOv8, etc. The accuracy rate, recall rate, mAP50 and MAP50-95 of YOLOv10-X model reached 99.3%, 99.9%, 99.5% and 97.4% respectively, but YOLOv10-S and YOLOv10-N had the fastest training time and detection speed, and were more suitable for mobile intelligent devices.