Vietnam National University, Hanoi

Microplastics (MPs), defined as plastic particles smaller than 5 mm, originate from the degradation of larger plastic items or manufactured products, such as microbeads. Stormwater runoff is increasingly recognized as a critical pathway for introducing MPs into marine environments, thereby threatening both ecosystems and human health. Despite growing concerns over MPs as emerging contaminants, a significant knowledge gap remains regarding their global distribution and the associated challenges to human health. This highlights the need for comprehensive research into the factors influencing MP contamination in stormwater runoff. To collect data, the search was restricted to academic articles published between 2020 and 2024. After careful screening, 45 reports were selected and included in the current study. The sizes and global levels of MPs in stormwater runoff vary widely, with dominant particles being less than 1000 µm and concentrations ranging from as low as 1.2–3 particles/L to as high as 11,932 ± 151 to 18,966 ± 191 particles/L. Common MPs identified in stormwater include fragments, fibers, foams, films, rubbery particles, and spheres. The research findings underscore stormwater runoff as a significant pathway for transporting MPs within urban settings. Therefore, it is essential to prioritize treatment strategies specifically designed to remove MPs effectively. This necessitates adopting nature-based solutions like bioretention cells, rain gardens, sustainable urban drainage systems (SUDS), and constructed wetlands. Such strategies harness the inherent capabilities of natural ecosystems to efficiently filter and capture MPs, thereby significantly reducing pollution from stormwater runoff.

This paper has two primary objectives. First, it contributes to the literature on oil stabilisation funds and price controls by examining how such a fund is used to regulate market prices in the developing country of Vietnam. Second, it employs descriptive statistics and a standard GARCH methodology to investigate whether the fund, which operates as a form of price control, can effectively reduce domestic price volatility. The results show that the oil price stabilisation fund failed to achieve its intended goal. Considering the administrative costs and other negative impacts associated with the fund, a more market‐oriented approach, potentially combined with a price‐elastic tax system, is recommended for determining domestic oil prices.

Merging multi-source precipitation data based on deep learning models to create an accurate rainfall dataset has received significant interest in recent years. This article proposes a deep learning model to produce a high-accuracy, near real-time precipitation product for the North Central region of Vietnam during the period 2019–2023, with a spatial resolution of 0.04 ⁰ and a temporal resolution of 1 hour. The input multi-source data including near real-time satellite-derived precipitation products (PERSIANN-CCS, GSMaP-NRT, and IMERG-Early Run), radar precipitation, and gauge observations, and spatial features NE and POP are merged by a multiscale CNN based model with focal loss function and mean square error loss function for classification and regression tasks, respectively. Extensive experiments demonstrate that the proposed precipitation product outperforms all the input precipitation products and the post-real-time global precipitation products including GSMaP-MVK-Gauge and IMERG-Final Run. It achieves classification metrics with a CSI of 0.65 and a BIAS of 1.03, with improvements from 31.58% to 54.8% in CSI and from 17.47% to 105.82% in BIAS, compared to radar, GSMaP-MVK-Gauge, and IMERG-Final Run products. For regression metrics, it achieves an RMSE of 3.34 mm/h, and an mKGE of 0.70, with improvements from 10.18% to 100% in RMSE and from 15.71% to 71.43% in mKGE over the same reference products. These results indicate that the merged product has a greater capability to detect rainfall events and significantly better overall performance, with lower systematic and random errors compared to the same reference products. Moreover, the proposed method outperforms the other methods, including Random Forest, Long Short-Term Memory, and the original multiscale CNN.

Transcatheter aortic valve replacement has emerged as a valuable alternative to surgical aortic valve replacement in patients with severe aortic stenosis. Given the expansion of transcatheter aortic valve replacement to lower‐risk and younger populations with longer life expectancy, the durability of transcatheter heart valves (THVs) has become an important issue that may impact cardiovascular outcomes. THVs share similarities with surgical valves but have unique features, including a trend to larger effective orifice area and less prosthesis–patient mismatch, interactions with the native valve, and crimping process, that may all potentially influence a THV's life span. Multiple mechanisms may lead to bioprosthetic valve dysfunction, including structural valve deterioration, thrombosis, endocarditis, and nonstructural valve deterioration. With an incidence of up to 12.3% 5 years after transcatheter aortic valve replacement, structural valve deterioration represents the ultimate consequence of fibrotic remodeling and calcification within the bioprosthesis, driven by thrombotic and inflammatory processes involving the native aortic valve and influenced by patient and procedural factors. Understanding these mechanisms is crucial for improving THV durability.

Composites reinforced by carbon nanotubes (CNTs) with extraordinary properties are increasingly used to improve the performance of structures under complex dynamic and thermal loading. This paper studies the transient response to a moving mass of inclined composite microbeams reinforced by CNTs, focusing on the influence of CNT agglomeration. The effective properties of the composite are temperature-dependent, and they are estimated by the Eshelby-Mori–Tanaka approach. Considering both the shear deformation and rotary inertia, a size-dependent finite element beam model is derived in the framework of the n-order shear deformation theory and the modified couple stress theory (MCST). The transverse shear rotation rather than the sectional rotation is adopted as an independent variable in the model which helps to fulfill a longitudinally linear variation of the bending strain. The transient response is predicted for a simply supported microbeam with different inclination angles and CNT volume fractions. The result reveals that the agglomeration reduces efficiency of the CNT reinforcement, and the increase of CNT volume fraction does not improve the transient response when the agglomeration is severe. The effect of temperature rise on the transient response is found to be governed by the degree of CNT agglomeration and the microstructural size effect. The effect of temperature rise is more pronounced when the CNT agglomeration is severe, while this effect becomes insignificant for the microbeam associated with a higher size scale parameter. The influence of the CNT reinforcement, inclination angle, and the size scale parameter on the thermoelastic transient behavior of the microbeams is studied in detail.

Abandonment of agricultural land during urban development reduces land use efficiency, land quality, and urban aesthetics. It is affected by many different factors. Until now, there has been no research to determine all the factors influencing agricultural land abandonment and their influence levels. So, the study's primary purpose is to evaluate the current state of agricultural land abandonment and the level of simultaneous impact of different factors on it during urban development. The study randomly investigated 255 households abandoning agricultural land about the factors causing that phenomenon by two steps. The hypothesized model of factors was tested through evaluation criteria. Research results have shown for the first time that agricultural land abandonment is affected by 7 factor groups. Their impact rate ranges from 5.21% to 26.61%. The land factor group has the most substantial impact; Officials and civil servants have the slightest effect. Proposed solutions to reduce agricultural land abandonment and use it more efficiently include converting low-yield agricultural land to non-agricultural land, ensuring infrastructure for agricultural land use, providing financial support, livestock, and seedlings to farmers during production, and encouraging farmers to rent land when they are not using it.

Introduction Malignant hyperthermia (MH) is a fatal hypermetabolic reaction of skeletal muscle, triggered by exposure to volatile anesthetic agents or depolarizing muscle relaxants. It typically exhibits hypercarbia, muscle rigidity, tachycardia, and hyperthermia. Diagnosis is often confirmed through a muscle biopsy for the in vitro contracture test or by identifying pathogenic genetic variants. Case Presentation We report 2 cases of suspected MH. The first case involved a 4-year-old female (20 kg) undergoing adenotonsillectomy, and the second involved a 13-year-old female (56 kg) who underwent pedicle screw fixation surgery. Both patients had unremarkable medical histories. During maintenance of general anesthesia with sevoflurane, they developed clinical signs highly suggestive of MH—10 minutes after exposure in the first case and 120 minutes after exposure in the second case. Both cases were managed with dantrolene and supportive care. In the first case, dantrolene was administered 4 hours after the initial signs, by which time significant rhabdomyolysis had already developed. In the second case, early administration within 10 minutes was associated with a much milder degree of rhabdomyolysis. Conclusion Early recognition of symptoms and accurate differentiation of MH from similar conditions are essential for favorable outcomes. Prompt administration of dantrolene at the first sign of an MH reaction is critical for effective management.

Background Vietnam has one of the highest hepatitis B virus (HBV) infection rates, with approximately 8 million people affected. Although antiviral drug resistance mutations have been reported in treatment-naïve patients with chronic hepatitis B, there is limited data on primary drug resistance mutations in circulating genotypes within this population. Objectives This study aimed to investigate primary antiviral drug resistance mutations and common HBV genotypes in treatment-naïve patients with chronic hepatitis B, particularly in cases without well-characterized resistance profiles. Design A cross-sectional study. Methods We analyzed HBV genotypes and antiviral drug resistance mutations in 113 treatment-naïve patients with chronic hepatitis B in the Yenphong Medical Center, Bacninh Vietnam. The reverse transcriptase (RT) region of the HBV polymerase genes was sequenced to detect mutations. Results Genotypes B, C, and G were identified in 85.0% (96/133), 14.1% (16/133), and 0.9% (1/133) of treatment-naïve patients with chronic hepatitis B, respectively. Mutations in the RT region associated with antiviral drug resistance were detected in 32.7% (37/113) of patients. In addition, the most frequent resistance mutations were rtV207M (89.2%, 33/37), followed by A194T, L180M + M204V, V173L + M204I + L80I, and A181T + V207M + A181T, each observed in 2.7% (1/37). Notably, no significant associations were found between resistance mutations and HBV genotype, gender, age, hepatitis B e-antigen status, baseline HBV DNA levels, or level of alanine aminotransferase, aspartate aminotransferase, gamma-glutamyl transferase. Conclusion This study highlights the presence of primary resistance mutations in treatment-naïve patients and underscores the importance of genotypic screening prior to initiating therapy. These findings may inform treatment strategies and help reduce the risk of treatment failure, liver cirrhosis, and hepatocellular carcinoma.

This study investigates the low-dimensional dynamics of the wake flow around a blunt-based axisymmetric model under low-speed conditions. To capture the underlying flow structures, we apply various data-driven techniques, including proper orthogonal decomposition (POD), Hilbert-Huang transform (HHT), Hankel alternative view of Koopman (HAVOK), and cluster-based reduced-order modeling (CROM). Our findings reveal that vortex shedding, bubble pumping, and the stochastic rotation of vortex shedding are the three dominant modes of the wake flow, each associated with distinct Strouhal numbers. Further analysis of the first five POD mode coefficients using the Hilbert-Huang transform uncovers intermittent behavior in the near-wake dynamics, demonstrating the superiority of HHT over Fourier-based methods for characterizing such unsteady phenomena. The HAVOK analysis decomposes the delayed-embedding representation of the first POD mode into linear and nonlinear components, where the eigenvalues of the linear component capture very low-frequency oscillations and the bubble pumping mechanism. Meanwhile, CROM offers an unsupervised alternative to POD, effectively identifying transition dynamics and predictive information of cluster-based flow structures.

Image captioning is the detailed portrayal of the content of an image using natural language, which often requires a large amount of paired image-caption data during training. However, these pretrained models have difficulty adapting to new domains with novel objects that rarely or never appear during training. In this paper, we introduce the zero-shot novel object captioning task, which generates descriptions for novel objects without the need for paired image-caption data and extra training sentences. We propose a model called data key multitask learning conformer with representative mask for novel object image captioning (KMMC) with the following components: (1) introduction of a new multitask learning model based on data keys: data key multitask learning (DKMT). Unlike conventional multitask learning models that use decision heads, our model relies on (key, value) pairs to determine the execution of different tasks. This results in a lightweight model with scalability for multiple arbitrary tasks and maximum parameter sharing; (2) proposing the representative mask (RM) method to share context from seen objects for novel objects, generate multi-attention, and combine it with DKMT to create a robust triplet activation. This approach increases the likelihood of novel object labels appearing in the description compared to other placeholder methods. Our model has been tested on the COCO, NoCaps and ImageNet datasets, yielding competitive results with other SOTAs in novel object captioning.

Tumor necrosis factor‐alpha (TNF‐α) is a crucial cytokine that orchestrates inflammatory responses within the immune system. Derived from cells such as macrophages and monocytes, TNF‐α plays a pivotal role in inflammation by binding to its receptors on target cells. This binding initiates a cascade of events, including the production of other pro‐inflammatory cytokines and the recruitment of immune cells to the affected site. While TNF‐α is vital for the body's defense against infections and injuries, its sustained or excessive release can contribute to chronic inflammation and tissue damage. Remarkably, therapeutic interventions aimed at TNF‐α, such as marine compounds, have effectively managed inflammatory conditions like rheumatoid arthritis and inflammatory bowel diseases. Hence, the present study identifies natural compounds sourced from the Rhabdastrella providentiae sponge through in silico screening, involving molecular docking, drug‐likeness analysis, oral toxicity prediction, and density functional theory calculation. Docking simulation results reveal that rhabdastrellin G, rhabdaprovidines G, rhabdastrellin I, rhabdastrellin H and rhabdastrellin K exhibit stronger binding affinity than the reference inhibitor SPD‐304 (Δ G = −8.71 kcal/mol), with Δ G values of −9.583, −9.509, −9.877, −9.196, and −8.892 kcal/mol, respectively. Furthermore, drug‐likeness and oral toxicity analyses indicate that these compounds violate Lipinski's two rules but satisfy criteria for drug‐like natural compounds within the “known drug space” rules. Additionally, the predicted toxicity levels for rhabdastrellin G, rhabdastrellin H and rhabdastrellin K are lower and safer than those of other compounds under investigation. Therefore, rhabdastrellin G, rhabdastrellin H and rhabdastrellin K emerge as three potential candidates for further research in subsequent stages.

Extreme rainfall from August 4 to 6, 2023, severely impacted Ho Bon commune in Mu Cang Chai district, Yen Bai province. The event triggered a significant number of landslides and debris flows across the area. This study aims to investigate the conditions leading to shallow and large flow-like landslides in Ho Bon commune through field investigations, geological and geotechnical assessments, and statistical analysis of rainfall data. A total of 346 landslides were identified and categorized into four main types: six cases of rockfall, five cases of translational slide, 237 cases of debris flow, and 98 cases of complex landslide. Several significant characteristics were observed during this event, including (i) rainfall intensity ranking within the top 1% over the past 24 years (2000–2023); (ii) a series of shallow landslides with small to medium volumes, which occurred around 6:00–8:00 pm on August 5 and subsequently triggered large flow-like landslides in adjacent streams; and (iii) the influence of multiple contributing factors, such as intense weathering, fault zones, lineaments, high joint density, and the presence of silty sand with high hydraulic conductivity and low soil suction. In addition, human activities such as deforestation, rice paddy cultivation, and slope cutting significantly increased landslide susceptibility.

Corrosion detection in critical structural components is a persistent challenge in the aerospace and structural health monitoring industries. This work introduces an advanced Gaussian-Pulsed Eddy Current Testing (GPECT) system integrated with a Spiking Neural Network (SNN) for precise and efficient detection of corrosion. Unlike conventional square-pulse excitation, the proposed GPECT system employs a Gaussian pulse, enabling selection of the frequency bandwidth to enhance defect detection sensitivity. The system’s sensor probe features a coil for magnetic excitation and a Hall sensor centrally positioned to capture the resulting magnetic field signals. These signals are transformed into the frequency domain using Short-Time Fourier Transform (STFT), facilitating the extraction of key spectral features indicative of corrosion. Leveraging the temporal and energy-efficient processing capabilities of the SNN, which incorporates spike generation, convolutional feature extraction, and robust classification, the system achieves significant improvements in detectability and energy savings (10 time smaller than conventional NN model). Experimental evaluations on aluminum specimens with artificially induced corrosion of varying depths and diameters validate the system’s effectiveness, demonstrating high accuracy and robustness in differentiating corroded and non-corroded regions. This work establishes a robust foundation for next-generation nondestructive testing techniques, pushing the frontiers of corrosion detection in high-stakes applications.

Introduction This study aimed to evaluate the therapeutic effect of osimertinib and further to compare the results of osimertinib plus brain radiation vs. osimertinib monotherapy in advanced EGFR-mutant non-small cell lung cancer (NSCLC) patients with brain metastases (BMs). Methods A retrospective study was conducted involving 62 advanced EGFR-mutant NSCLC patients with BMs who were treated with first-line osimertinib at the Vietnam National Cancer Hospital between April 2019 and December 2023. Patients were categorised in two treatment groups: (1) osimertinib alone (33 patients) and (2) osimertinib combined with locoregional therapy, including stereotactic radiosurgery or whole-brain radiotherapy (29 patients). Endpoints included objective response rate (ORR), central nervous system response rate (CNS-ORR), progression-free survival (PFS), overall survival (OS). Results The systemic ORR was 91.9% and the disease-control rate (DCR) was 96.8%. The CNS-ORR was 91.9% and the CNS-DCR was 100%. The median PFS and median OS achieved were 24.5 and 35.2 months, respectively. There was no significant difference in outcomes between patients in either treatment group with respect to CNS-ORR ( P = 1.0), mean best percentage change from baseline in CNS target lesion size (P = .376), median PFS (P = .656), intracranial progression-free survival (iPFS) (P = .706), or OS (P = .734). The occurrence of any-grade adverse events (AEs) did not differ significantly between the two treatment groups (P = .762). However, in the osimertinib plus brain radiation cohort, 3/29 (10.3%) patients experienced radiotherapy-related AEs (2 cases of brain necrosis, 1 case of leukoencephalopathy), which consisted of one case of grade 3 brain radiation necrosis. Conclusion Osimertinib shows favorable real-world outcomes in improving PFS, OS, and CNS-ORR in advanced EGFR-mutant NSCLC Vietnamese patients with BMs, with no clear additional benefit from combining with brain radiotherapy.