Abstract: This paper introduces the concept of Existential Resistance Literature as an emerging interdisciplinary framework positioned at the intersection of philosophy, leadership theory, and socio-technical systems. The study responds to accelerating technological developments that increasingly frame human behavior through algorithmic, predictive, and data-driven models. While such systems enhance efficiency and coordination, they simultaneously risk reducing human agency, meaning, and interpretive depth.Building on a perception-centered perspective, the paper proposes that contemporary systems face a fundamental challenge: not merely optimization, but the preservation of human coherence. In response, Existential Resistance Literature is conceptualized as a human-centered intellectual and narrative approach that resists reductionist interpretations of human identity. Central to this framework is the concept of perceptual integrity, which explains how individuals and systems maintain meaning, trust, and continuity under conditions of complexity and technological mediation.The study integrates recent research on cognitive diversity, collective intelligence, and human–AI interaction to demonstrate that sustainable systems depend not only on structural efficiency but on interpretive alignment. By reframing resistance as a constructive process of preserving meaning rather than opposing technology, the paper advances a novel paradigm for understanding the relationship between human systems and algorithmic environments.

Abstract: The challenge of context-free validity arises from the common belief that rigorous methodology ensures research credibility in various contexts, despite variations in epistemic foundations, institutional capacity, cultural norms, and operational conditions. This assumption is clear in Global South contexts, where research tools and evaluation frameworks from other regions are applied without proper adaptation, highlighting the limitations of claims to universal validity. The challenge is especially evident in socioeconomic research, where tools and frameworks are often applied across contexts without accounting for institutional capacity, cultural norms, or resource limitations. This paper presents the Contextual Research Validity Index (CRVI), a framework for evaluating how well a research design fits the epistemic, institutional, cultural, and operational aspects of its intended context. The CRVI views contextual validity as a form of legitimacy, emphasising that a method’s credibility relies not only on technical precision but also on how well its assumptions align with the realities of the environment. The framework includes four dimensions—epistemic alignment, institutional fit, cultural resonance, and operational feasibility—combined into a composite index for systematic assessment. By focusing on contextual alignment, the CRVI addresses shortcomings in existing validity frameworks and provides researchers, evaluators, and practitioners with a tool to anticipate misfits, adapt designs, and enhance interpretive robustness. By redefining validity as a relational outcome and treating contextual coherence as a quantifiable aspect of rigour, the CRVI provides a systematic framework for assessing the legitimacy of research across diverse contexts.

Abstract: Fatigue is a frequent, disabling and difficult to treat symptom of multiple sclerosis (MS). Low grade inflammation and energetic dysfunction have been proposed as mechanisms underlying the pathogenesis of this symptom. Owing to its anti-inflammatory and metabolic properties, there is a rational for ketogenic diet (KD) application in this setting. We conducted a single arm open label interventional study on a strictly selected group of 16 non-obese patients with multiple sclerosis who were prescribed KD for three months. With respect to baseline, at 3 months we observed a significant reduction of fatigue severity scale (5.18 ± 1.02 vs. 4.16 ± 0.98; p = 0.042), Epworth Sleepiness Scale (5.64 ± 2.46 vs. 8.46 ± 3.05; p < 0.001), Pittsburgh Sleep Quality Index (5.64 ± 3.53 vs. 7.62 ± 2.59; p = 0.009), Depression Anxiety Stress Scales-21 depression (3.18 ± 2.93 vs. 6.15 ± 3.81; p = 0.036) and anxiety (5.15 ± 4.10 vs. 1.55 ± 1.92; p = 0.019) sub-scales, and an improvement in energy sub-scale of Multiple Sclerosis Quality of Life-54 (52.49 ± 12.83 vs. 37.43 ± 14.26; p = 0.042). These findings suggest that KD might be useful for the treatment of fatigue and they raise the interest for the use of KD in the treatment of other symptoms frequently encountered in multiple sclerosis.

Abstract: In cluster-merger analyses, the dominant gravitating component is often modeled as effectively history-independent after several dynamical times, even if the gas retains thermodynamic signatures of past perturbations. Recent weak-lensing work by HyeongHan et al. (2025) complicates that expectation for the Perseus Cluster by reporting a massive sub-halo, centered on NGC 1264, and a connecting mass bridge in a cool-core system long treated as a benchmark relaxed cluster. Perseus is already known from X-ray studies to host large-scale sloshing and an ancient cold front that preserve evidence of past perturbation on Gyr (gigayear) timescales. Taken together, these results motivate a re-examination of how merger history can remain observationally relevant in nominally relaxed clusters. This paper advances a deliberately modest claim. Rather than treating Perseus as a standalone falsification of ΛCDM or of conventional hydrodynamical explanations, this paper treats it as an especially informative case in which a remnant stress-energy interpretation becomes interesting enough to warrant further study. In this interpretation, long-lived gravitational structure is represented phenomenologically by a coarse-grained remnant stress-energy TμνRem, motivated by a covariant closure construction. The principal contribution of the paper is a falsifiable observational program rather than a claim of proof. After controlling for instantaneous merger parameters, residual lensing-gas centroid offsets in nominally relaxed clusters should correlate with independent merger-history proxies if such remnants are physically relevant. Existing lensing and X-ray archives already permit a pilot test, while upcoming wide-field surveys can extend the sample.

Abstract: Ordinary differential equations are fundamental tools for modeling dynamic systems in science, engineering, and applied mathematics. Solving these equations accurately and efficiently is crucial, particularly in cases where analytical solutions are challenging or impossible to obtain. This paper presents a method for solving inhomogeneous linear ordinary differential equations using an artificial neural network. The network is composed of a single input layer with one neuron, one hidden layer with three neurons, and a single output layer with one neuron. A multiple regression model is employed to determine the weights from the input layer to the hidden layer, while radial basis functions are used to compute the weights from the hidden layer to the output layer. The bias values are chosen within the range of -1 to 1 to optimize learning behavior. A trial solution is constructed as a sum of two parts. One part satisfies the initial condition, and the other part is the output of the network to approximate the function. The neural network is trained to minimize the mean squared error of the residuals obtained by doing the substitution of the trial solution into the given ordinary differential equation. The methodology is tested on first-order and second-order ordinary differential equations to evaluate its accuracy, stability and how its capability can be generalized. The results show that the method can approximate the exact solutions of these ordinary differential equations with high accuracy.

Abstract: Background/Objectives: WHO has identified Carbapenem-Resistant Acinetobacter baumannii (CRAb) and carbapenem-producing Enterobacterales (CPE) as the “critical priority” group of MDR organisms for which new therapeutic strategies are urgently needed. Here, we evaluated in vitro synergistic activity of eugenol, cinnamaldehyde and carvacrol in combination with β-lactams, gentamicin, or colistin against multidrug-resistant (MDR) Gram-negative bacteria (GNB). Methods: We selected seven MDR-GNB clinical isolates inclining CRAb, ESBL-producing and CPE clinical isolates displaying different antimicrobial susceptibility profiles. The genomes of clinical isolates were characterized by whole-genome sequencing and synergy testing was performed with checkerboard assay. Results: Our results demonstrated that eugenol, cinnamaldehyde and carvacrol in combination with colistin exhibited high synergistic activity against MDR-GNB clinical isolates (37.5-50%), while the effect was almost indifferent in combination with different β-lactam molecules or gentamicin (87.5-100%). Synergistic interaction of eugenol, cinnamaldehyde and carvacrol with colistin induced a statistically significant reduction (p<0.05) of the MIC values compared with the molecules tested alone. Conclusions: Our data demonstrated showed that this synergistic interaction was not affected by different antimicrobial resistance genes and/or different antimicrobial susceptibility profiles. In conclusion, our results suggest that eugenol, cinnamaldehyde and carvacrol in combination with colistin represents a potential strategy for treatment of MDR-GNB pathogens and limit their diffusion.

Abstract: Artificial Intelligence of Things (AIoT) applications increasingly exceed the limits of centralized cloud processing because they require low latency, privacy preservation, scalability, and operational resilience. This review synthesizes distributed intelligence across the edge–fog–cloud continuum through a structured integrative methodology comprising multi-stage literature search, two-stage filtering, and thematic synthesis of more than 100 sources. The analysis covers four representative domains—industrial IoT, smart cities, connected healthcare, and smart agriculture—to identify recurring architectural patterns and shared deployment challenges. The review organizes these challenges around power and computational constraints, data management, security and privacy, interoperability, and model lifecycle management. Building on this synthesis, the paper formalizes an Edge–Fog–Cloud distributed intelligence model and develops a workload-placement taxonomy based on latency, privacy, power, and model complexity. Comparative analysis shows that on-device TinyML is best suited to ultra-low-latency and privacy-sensitive inference, edge and fog layers provide an effective compromise for localized near-real-time intelligence, and cloud infrastructures remain essential for large-scale analytics and model training. Across domains, the evidence supports hybrid multi-layer architectures as the most robust strategy for advanced AIoT deployments. The review also identifies key future directions, including human-in-the-loop AIoT, multimodal sensor fusion, energy-harvesting devices, federated learning, and the Tactile Internet.

Abstract:

Chordoma is a rare malignant neoplasm of the axial skeleton, arising from notochordal remnants. No approved systemic therapies exist and a 10-year overall survival is below 60%. Accurate molecular and pathological classification is a prerequisite for improved prognostication and identification of actionable therapeutic targets, yet molecular classification of chordoma remains significantly less advanced than in other neoplasms. This article reviews and synthesizes proposed classification frameworks for chordoma across histological, radiological, surgical, genomic, epigenomic, transcriptomic, and proteomic domains. PubMed and CENTRAL were searched on 1 February 2026 using five queries: ‘chordoma classification’, ‘chordoma DNA sequencing’, ‘chordoma RNA sequencing’, ‘chordoma methylation’, and ‘chordoma copy number’. Original research articles describing more than one patient and reporting a classification or subtyping framework were included; review articles, case reports, and non-English publications were excluded. Sample size and utilization of validation dataset were identified for each dataset to mitigate risk of bias. Results were synthesized qualitatively. 108 studies encompassing 6,349 individuals were included. Across six domains, four cross-cutting themes with prognostic and potential theranostic value emerged: copy number alterations — particularly CDKN2A/B loss; SWI/SNF complex dysfunction; TGF-β signaling; and immune microenvironment heterogeneity.

Abstract: Background: Posterior segment eye diseases, including age-related macular degeneration and diabetic retinopathy, are preeminent causes of vision loss worldwide. Effective drug delivery to the retina poses an ongoing therapeutic difficulty due to the presence of the anatomical and physiological barriers. Nanotechnology-based drug delivery systems represent a promising strategy to overcome those limitations. Methods: A narrative literature review was conducted using the PubMed, Scopus, and Google Scholar databases, covering publications published between 2020 and 2026. Publications evaluating nanoparticles for the treatment of the vitreoretinal disorders, including pre-clinical in vitro and in vivo studies, were analyzed. Results: Nanocarriers, including liposomes, polymeric nanoparticles, and lipid-based systems, established improved drug bioavailability, stability, and targeted delivery. The analyzed systems facilitate sustained drug release and potentially reduce the prevalence of invasive intravitreal injections. The nanocarriers’ effectiveness is primarily influenced by their physicochemical properties, such as particle size, surface charge, and encapsulation efficiency. Nonetheless, the production costs and safety aspects, including cytotoxicity, oxidative stress, and inflammatory responses, remain as significant limitations. Conclusions: Nanotechnology-based drug delivery systems serve as an auspicious therapeutic approach for posterior segment eye diseases. However, further standardized preclinical and clinical research is required to assure long-term safety, and enable successful clinical transition.

Abstract: Autism spectrum disorder (ASD) frequently manifests with profound language impairments, particularly in verb morphology processing, which hinges on fronto-temporal connectivity for grammatical rule application. This study pioneers the use of graph neural networks (GNNs) to map these deficits, analysing task-based fMRI data from 72 children (36 ASD, 36 controls). Fronto-temporal graphs were constructed with nodes representing key regions (e.g., inferior frontal gyrus, superior temporal gyrus) and edges capturing dynamic Pearson correlations during an auditory verb tense judgment task. A three-layer GraphSAGE model, incorporating message passing and temporal embeddings, achieved 91.7% classification accuracy (AUC=0.95), outperforming traditional classifiers by 14%. Attention maps revealed hypo-connectivity in the arcuate fasciculus pathway (p<0.001), correlating with ADOS language scores (r=-0.62), alongside compensatory frontal hyperconnectivity. Ablation studies confirmed the model’s reliance on task-evoked dynamics. These findings elucidate the neural substrates of morphology impairments, offering interpretable biomarkers for early ASD diagnosis and personalized interventions. By bridging graph theory with cognitive neuroscience, this work advances precision psychiatry, with implications for neurofeedback therapies targeting syntactic networks. Future extensions to multi-modal data promise enhanced generalizability across ASD heterogeneity.

Abstract: This study introduces the R-index as a novel framework for quantifying the economic impact of risk through realized deviations from expected performance. In contrast to traditional risk measures that rely on probabilistic or volatility-based approaches, the proposed index captures risk as an outcome-based phenomenon directly linked to firm-level performance. The R-index is constructed as a normalized measure of deviation between actual and expected values and is further extended to a multidimensional setting, allowing for aggregation across different performance indicators. The empirical analysis is conducted using longitudinal financial data from three firms operating in distinct sectors of the Montenegrin economy—telecommunications, retail, and tourism—over the period 2015–2024. The results reveal substantial heterogeneity in the realization of risk across firms, even under identical macroeconomic conditions. While some firms exhibit stable performance and limited deviations, others demonstrate pronounced volatility and sensitivity to external shocks, particularly during the COVID-19 period. These findings suggest that risk is not uniformly transmitted but is instead shaped by firm-specific characteristics, including operational structure and adaptive capacity. The study contributes to the literature by redefining risk as a realized economic phenomenon and by proposing a scalable and interpretable metric that bridges risk measurement and performance evaluation. The R-index offers practical relevance for managerial decision-making and provides a foundation for future research on the relationship between risk and firm value.

Abstract: This study investigates how Gen Z’s digital engagement with the Diyandi Festival in Iligan City reconfigures cultural participation into a hybrid experience—where faith, identity, and storytelling unfold across both physical and digital spaces. Employing a qualitative case study design, enriched by digital ethnographic and content analytic methods, the research draws from semi-structured interviews, written and online artifacts, and a robust theoretical framework that includes Hall’s encoding/decoding model, Bakhtin’s carnivalesque theory, and Adorno’s critique of mass culture. Twenty salient themes emerged, mirroring the voices, views, sentiments, and lived experiences of young Iliganons as they navigate tradition through memes, livestreams, and remix aesthetics. Overall, the paper encapsulates the symbolic tension between sacred ritual and digital disruption, highlighting the fragility of mediated spirituality. The study’s innovation lies in its fusion of ethnographic depth with digital cultural analysis, offering a localized yet globally resonant portrait of participatory heritage. It positions Iligan’s festival as a living archive that stands resilient against the tide of global cultural information. Limitations in scope and generational range prompt recommendations for comparative and longitudinal research, and suggest the viability of a “phygital” festival model—one that blends physical celebration with digital engagement to ensure cultural continuity in the age of information.

Abstract: Cardiovascular disease is traditionally interpreted through macrocirculatory parameters such as cardiac output, vascular resistance, and epicardial coronary anatomy. However, clinical outcomes frequently diverge from predictions based solely on these indices, particularly in syndromes such as heart failure with preserved ejection fraction (HFpEF), cardiogenic shock, and sepsis-associated myocardial dysfunction. Increasing evidence suggests that the integrity of the microvascular–immune interface plays a central role in determining tissue perfusion and cardiovascular resilience. This review proposes a staged framework of cardiovascular decompensation centered on progressive failure of this interface. In Stage 1, chronic cardiometabolic and inflammatory stress produces a primed but compensated microvascular state characterized by endothelial activation, glycocalyx vulnerability, pericyte remodeling, platelet sensitization, and reduced lymphatic reserve. Perfusion is preserved at rest, but vasodilatory reserve and microvascular stability are reduced, narrowing the effective perfusion window under physiologic stress. In Stage 2, acute insults such as infection, ischemia, or neurohumoral activation precipitate threshold instability within the microcirculation. Perfusion becomes governed by the arterial pressure–critical closing pressure (Pa − Pcrit) relationship rather than traditional arterial–venous gradients. As this window narrows, segmental capillary derecruitment and heterogeneous flow emerge, producing loss of hemodynamic coherence in which systemic blood pressure and cardiac output may appear preserved despite impaired tissue perfusion. In Stage 3, inflammatory amplification and immunothrombotic processes consolidate microvascular dysfunction. Pericyte contraction, endothelial injury, cytokine escalation, and neutrophil extracellular trap formation promote platelet–fibrin deposition and capillary obstruction, transforming reversible conductance failure into structural microvascular impairment. This framework provides a unifying physiologic lens for diverse cardiovascular syndromes, including Type 2 myocardial infarction, HFpEF decompensation, and cardiogenic shock. It also suggests that therapeutic efficacy may depend less on macrocirculatory normalization alone and more on preserving microvascular integrity before immunothrombotic consolidation occurs. Although this model remains hypothesis-generating, it highlights the microvascular–immune interface as a central determinant of cardiovascular stability and a potential target for future precision hemodynamic and immunomodulatory strategies.

Abstract: Background: The expansion of online and hybrid graduate education has shifted the central quality question from delivery feasibility to whether institutions can credibly demonstrate advanced, assessable graduate capability in digitally mediated environments. Competency-based education offers a promising framework for this challenge, but its conceptual foundations and implementation logics remain uneven across higher education. Objective: This scoping review maps how competency-based curriculum design is conceptualised and operationalised in online graduate education and derives context-sensitive implications for emerging African universities. Methods: Guided by Joanna Briggs Institute scoping review methodology and a Population-Concept-Context framework, the review synthesised peer-reviewed studies alongside selected policy and quality assurance documents relevant to online graduate education, competency-based design, and digital higher education governance. The analysis was interpreted through Constructive Alignment, Community of Inquiry, and TPACK. Results: The evidence converged around six interdependent domains: competency specification, curriculum architecture, assessment evidence chains, online interaction design, learning management system configuration, and faculty and governance capability. The review found that the central problem is not merely definitional ambiguity, but the failure to sustain alignment from competency statements to valid assessment, platform workflows, and institutional quality assurance. It also found that much of the available evidence comes from higher-capacity systems and professionally regulated disciplines, limiting direct transferability to emerging African universities. Conclusion: Competency-based online graduate curricula are most defensible when treated as institution-wide design architectures rather than course-level innovations. For emerging African universities, credible implementation depends on coherent alignment among curriculum, pedagogy, assessment, platform design, faculty development, and quality management. The review therefore argues for selective translation rather than uncritical borrowing of dominant models.

Abstract: We develop a queueing-organized framework for within-venue monitoring of BTC/USDT liquidity, signed-flow pressure, and resiliency on Binance. The model treats latent buy and sell pressure as occupancy processes and uses that state space to organize three empirical diagnostics: the variance-per-BTC liquidity measure R_r, the effective mean-reversion rate θ_eff, and the companion signed-flow proxy beta^proxy_eff. Using Binance trade data from 2020 to 2025, we find a pooled first-order variance-volume regularity away from the highest-volume tail and substantial time variation in rolling liquidity and resiliency. In overlapping 30-day windows, θ_eff is positive by point estimate in roughly two-thirds of windows but clearly positive in only about two-fifths under a simple uncertainty buffer, implying that local recovery is often fragile or ambiguous. The intended users are short-horizon risk managers, execution desks, market makers, and exchange surveillance teams that need auditable venue-level indicators of when liquidity is thinning, recovery is weakening, and signed flow is turning one-sided. Queueing is useful here because it turns those signals into one coherent monitoring dashboard for venue-level market quality and short-horizon risk.

Abstract: Supplemental oxygen is a cornerstone intervention in modern clinical practice, widely used to correct hypoxemia in emergencies, perioperative, and critical care settings. While oxygen therapy is lifesaving, accumulating evidence indicates that excessive oxygen exposure can induce significant pathophysiological disturbances, particularly within the cardiovascular and pulmonary systems. Hyperoxia (PaO₂ > 100 mm Hg) promotes the generation of reactive oxygen species (ROS), leading to oxidative stress, mitochondrial dysfunction, and activation of pro-fibrotic pathways. When combined with mechanical ventilation, these effects are further amplified through alterations in intrathoracic pressure, reduced venous return, and increased pulmonary vascular resistance, collectively imposing hemodynamic stress on the myocardium. These mechanical and biochemical perturbations converge to drive structural, functional, and electrical remodeling of the heart, including conduction abnormalities and arrhythmogenesis. Emerging clinical insights, particularly from critically ill and COVID-19 populations, underscore the importance of titrated oxygen strategies that balance adequate tissue oxygenation with minimization of hyperoxic injury. This review synthesizes current evidence on hyperoxia-induced oxidative stress, heart–lung interactions, and mechanisms underlying myocardial remodeling to provide a comprehensive framework for optimizing oxygen therapy.

Abstract: 4-Dimethylamino-2’-hydroxy chalcone (DHC) 1 is an important natural compound that is nearly non-fluorescent in solution but highly fluorescent in its crystalline state. At room temperature, the weak fluorescence from DHC solution is exclusively from its keto tautomer, without notable contribution from its enol tautomer. By using low temperature fluorescence, the study found that the enol emission could be detected upon cooling with liquid N₂ in a protic solvent (e.g. EtOH). This led to observation of the fluorescence vibronic structure of enol tautomer, in addition to its enol emission λ_em ≈ 473 nm that is well separated from its keto tautomer emission (λ_em ≈ 600 nm). By freezing DHC in a solvent matrix, the study revealed the fluorescent characteristics of a single molecule in a rigid environment. Further comparison of DHC in a solvent matrix and crystalline state disclosed that the emission of crystalline DHC was primarily from the keto tautomer, along with some minor contribution from the enol tautomer, despite the tight packing environment in the crystalline state.

Abstract: Background/Objectives: To analyse the causes, characteristics, and outcomes of hospital admissions in patients with systemic lupus erythematosus (SLE) over a 30-year period in a tertiary referral centre in Catalonia, and to evaluate temporal trends and prognostic factors associated with adverse outcomes. Methods: A retrospective observational study was conducted including all SLE patients admitted to the Department of Autoimmune Diseases at Hospital Clínic de Barcelona between June 1995 and December 2024. Admissions lasting less than 48 hours or lacking clinical documentation were excluded. Variables analysed included demographics, disease duration, comorbidities, cause of admission, treatments, and outcomes. A composite outcome was defined as intensive care unit (ICU) admission, 30-day readmission, or prolonged hospital stay. Statistical analyses included univariate and multivariate regression models. Results: Among the 1,216 hospital admissions, SLE flares and infections were the most frequent causes. Over the study period, admissions due to infections increased significantly and, in the last five years, exceeded those related to disease flares (33.7% vs. 26.1%). Patients hospitalized for flares were younger and had a shorter disease duration, whereas infection-related admissions were more common among older patients, those with overlap syndromes, and those with higher damage scores. Vascular events and SLE flares were independently associated with poorer outcomes. Although antimalarial use increased over time, it remained suboptimal, largely due to drug toxicity and newly diagnosed cases (from 45.2% to 69.7%; p< 0.001). Treatment strategies also evolved, with a shift toward lower gluco-corticoid doses (from 14.5% to 38.3%; p< 0.001), and mycophenolate mofetil replacing cyclophosphamide as the preferred immunosuppressive agent. Conclusions: Hospitalisation patterns in SLE have shifted over time, with infections emerging as the leading cause of admission. This trend reflects an evolving patient profile characterized by older age, greater accumulated damage, comorbidities, and increased exposure to immunosuppressive therapies. These findings underscore the need for optimized infection prevention strategies and individualized treatment approaches to improve outcomes in contemporary SLE care.

Abstract: Text classification in low-resource languages has become increasingly important due to the rapid growth of user-generated digital content. While multitask learning has long been studied in NLP, the use of LLMs for multitask text classification in low-resource languages such as Bengali remains underexplored. Although LLMs are inherently multilingual and multitasking, their effectiveness in structured multitask classification settings for Bengali has not been systematically evaluated. In this work, we investigate how LLMs can be leveraged for multitask Bengali text classification across five domains: sentiment analysis, aggressive text detection, fake news detection, news categorization, and emotion analysis. We compare in-context learning strategies—including zero-shot, one-shot, and chain-of-thought prompting—with parameter-efficient fine-tuning approaches. Our findings show that CoT prompting does not consistently improve performance and often degrades performance, highlighting the instability of prompt-based adaptation in low-resource settings with limited pretraining exposure. Moreover, reasoning-optimized models such as DeepSeek-R1 exhibit substantial performance drops, indicating that enhanced reasoning capabilities alone cannot overcome the challenges posed by low-resource settings. Among the evaluated mLLMs, Gemma-3-4B demonstrates the most stable and balanced cross-task performance under both in-context learning and parameter-efficient fine-tuning, making it a strong backbone candidate for multitask Bengali text classification. These results provide empirical evidence on the limitations of prompting and the advantages of lightweight fine-tuning for low-resource multilingual NLP.

Abstract: Dengue infection remains a major global health concern, with a subset of patients progressing from self-limited dengue fever to severe disease characterized by plasma leakage, shock, and organ dysfunction. The dengue non-structural protein 1 (NS1), a multifunctional glycoprotein expressed on infected cells and secreted into circulation, has emerged as a key mediator linking viral infection to immune-driven vascular pathology. This review synthesizes experimental, animal, and human clinical evidence on NS1-driven immunopathogenesis, focusing on mechanisms leading to endothelial dysfunction and increased vascular permeability. NS1 modulates the complement system in a context-dependent manner, contributing to immune evasion by inhibiting terminal complement complex formation, while also promoting antibody-dependent complement activation associated with severe disease. Additionally, NS1 directly disrupts endothelial barrier integrity through disruption of adherens and tight junction architecture, Ang-2/Tie2 imbalance, activation of RhoA/ROCK signalling, and enzymatic degradation of the endothelial glycocalyx, with further amplification through inflammatory mediators. In addition, evidence is integrated showing that NS1 activates innate immune signaling, perturbs platelet biology and haemostasis, forms pro-inflammatory complexes with lipoproteins. Moreover, anti-NS1 antibodies may be both protective and pathogenic. Collectively, these data position NS1-linked pathways as rational targets for adjunctive therapies and next-generation vaccines aimed at preventing vascular leakage and severe dengue infection.