Abstract:

Growing evidence for dynamical dark energy challenges the passive cosmological constant paradigm. This perspective article introduces a novel conceptual framework and a minimal, testable benchmark model to probe a fundamental question: is dark energy’s evolution correlated with cosmic structure growth, suggesting it is an intrinsic component of cosmic dynamics rather than a static background? We propose a linear correlation of the form w(a)=−1+η(γ(a)−0.55) between the dark energy equation of state w(a) and the structure growth index γ(a) as a key observational signature of this intrinsic link. This linear relation is the first concrete, testable benchmark framed from the perspective of dark energy as an intrinsic cosmic dynamical component. To provide physical motivation and verify self-consistency, we construct a phenomenological “Dynamic Coupling Model.” In this model, the energy transfer rate between dark energy and dark matter is postulated to be dynamically modulated by cosmic structure growth (traced by γ(a)). This model naturally yields the linear w-γ relation, with a theoretically motivated benchmark slope η=0.25±0.03. The model’s key testable prediction is a deviation at redshift z≈0.5, where w≈−0.89±0.02, in stark contrast to ΛCDM’s w=−1, offering a clear observational target. Future high-precision data will first verify the existence of this correlation. If confirmed, data can further discriminate whether it supports this simple linear parameterization or points to more complex coupling mechanisms. Regardless of the outcome, this w-γ correlation paradigm provides a new, actionable starting point for understanding dark energy’s dynamical role. The proposed framework is consistent with current cosmological data, shows potential to alleviate the Hubble tension, and defines a clear path for observational testing.

Abstract: Background/Objectives: Influenza B virus contributes substantially to annual morbidity and mortality, accounting for 20% to 30% of all influenza-associated deaths globally. Vaccination prevents severe disease, yet widely used inactivated influenza vaccines fail to reduce virus transmission. Also, influenza B viruses are less susceptible to commonly used antivirals than influenza A viruses. Therefore, new approaches are needed to decrease disease burden and limit virus spread. Neomycin, an aminoglycoside antibiotic, has recently been shown to mitigate SARS-CoV-2 transmission in a hamster model. In this study, we investigated the impact of neomycin on the transmission of influenza B virus. Methods: We used an influenza contact transmission model in guinea pigs and an aerosol transmission model in ferrets. Animals in experimental groups received intranasal neomycin sulphate (5 mg/guinea pig, 20 mg/ferret) or placebo one day before contact with infected animals and for four days thereafter. In the guinea pig experiment, an additional control group of animals was treated intranasally with interferon alpha. The virus spread from infected to contact animals was assessed by RT-PCR and viral culture of nasal washes during two weeks. Clinical signs and body weight were monitored daily. Results: In the guinea pig model, 75% of contact animals became infected with influenza B virus regardless of treatment. Neither neomycin nor interferon alpha prevented infection, although both delayed the onset of viral shedding in contact animals. In the ferret model, 33% of contact animals in the placebo group became infected, whereas no viral shedding was detected in the neomycin-treated group. Conclusions: Prophylactic intranasal neomycin treatment has the potential to protect exposed subjects from aerosol transmission of influenza B virus during flu outbreaks.

Abstract: Dog assisted interventions (DAIs) are an established procedure to support military staff but their implementation during active warfare has not yet been systematically studied. Alongside, the welfare of therapy dogs participating in DAIs during war remains unexplored. Therapy dogs may develop clinically relevant emotional disorders, including trauma-related stress responses, analogous to human psychopathologies. The present study sought to monitor physiological arousal in therapy dogs performing DAI sessions with their handlers in two Ukrainian military hospitals (Vinnyzja and Kyiv). Thus, biomarkers of hypothalamic-pituitary-adrenal (HPA) axis activity including salivary, urinary and hair cortisol concentrations in Ukrainian (UA) therapy dogs were assessed to capture their acute and long-term stress responses. Moreover, cortisol levels from German (GE) therapy dogs, performing similar DAIs under peaceful conditions were gathered to compare cortisol levels between dogs. Results suggest that GE dogs exhibited significantly higher urinary and hair cortisol levels and significantly lower salivary cortisol concentrations, reflecting alterations in longer-term glucocorticoid secretion that is possibly caused by war-related stimulation in the UA cohort. In contrast, no significant differences in salivary cortisol emerged as a consequence of performing DAIs. The present findings suggest an environmental impact on therapy dogs’ cortisol secretion rather than involvement in DAIs.

Abstract: Background: Pressure injuries (PIs) in surgical settings is a major patient safety concern, contributing to increased morbidity, prolonged hospitalization, and greater healthcare costs. Prolonged immobility, anesthesia, surgical positioning, and device-related pressure place perioperative patients at particularly high risk. This mini review synthesizes contemporary literature on PIs prevention and risk assessment in perioperative environments, with emphasis on the use of specialized tools, particularly the Munro Scale and evidence-based strategies such as optimized support surfaces, positioning, moisture management, and early skin assessment. Comparative findings indicate that the Munro Scale offers superior predictive accuracy for surgical patients compared with the widely used Braden Scale. The review also highlights persistent challenges, including limited implementation, lack of standardization, and insufficient knowledge among perioperative staff. A multidisciplinary, proactive, and individualized approach is essential to effective PI prevention, while future work should prioritize EHR-integrated risk tools, enhanced staff training, and broader adoption of periopera-tive prevention bundles.

Abstract: Oxidative and nitrosative stress are central mechanisms in the pathogenesis of neurodegenerative diseases, where excessive production of reactive oxygen and nitrogen species (ROS/RNS) leads to mitochondrial dysfunction, membrane damage, and neuronal death. In this study, we established and compared short-term (2 h) and long-term (20 h) exposure paradigms to sodium nitroprusside (SNP), used as a xenobiotic nitric oxide donor, in two neuronal cell lines (mHippoE-18 and PC12) and zebrafish larvae, aiming to provide a preclinical framework for neurodegenerative drug discovery. In vitro, SNP exposure caused concentration-dependent reductions in viability and alterations in oxidative balance, with mHippoE-18 cells exhibiting higher susceptibility than PC12 cells. In the short-term exposure, cytotoxicity was primarily associated with membrane disruption at higher concentrations, while oxidative stress contributed more strongly at intermediate doses. In the long-term exposure, mHippoE-18 cells showed strong integrated correlations between ROS, LDH release, and viability loss, highlighting their vulnerability to nitrosative stress. In zebrafish, SNP exposure impaired metabolic activity and swimming behavior in both paradigms. Long-term exposure led to consistent dose-dependent increases in ROS, accompanied by locomotor deficits tightly linked to energy metabolism. Overall, the higher sensitivity of mHippoE-18 cells compared with PC12, and the dose-dependent metabolic and behavioral impairments observed in zebrafish, indicate that cellular responses partially mirror the in vivo outcomes. This integrative approach underscores the value of combining neuronal cell lines with zebrafish larvae to capture complementary aspects of SNP-induced neurotoxicity and to strengthen preclinical evaluation of candidate compounds with protective or therapeutic potential. These findings support the use of SNP as a xenobiotic model to probe nitrosative stress–driven neurotoxicity across cellular and organismal systems.

Abstract: The overlapping circulation of influenza (Flu), severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; SC2), and respiratory syncytial virus (RSV) continues to challenge clinical laboratories, particularly in settings with limited automation and fragmented healthcare coverage. This study expanded the CDC Flu-SC2 assay by incorporating a laboratory-developed test (LDT) for RSV A/B detection into a fully automated quadruplex RT-qPCR (LDRA) on the Panther Fusion® Open Access™ system. The design, based on more than 8,000 RSV genomic sequences targeting the conserved M gene, achieved optimal amplification efficiencies (97–105%) and full multiplex compatibility. Analytical assessment established limits of detection between 9.6 and 37.8 copies per reaction, absence of cross-reactivity with 30 respiratory pathogens, and inclusivity for 32 viral variants. Commutability and diagnostic performance among the LDRA, CE IVD-marked Allplex™ SARS-CoV-2/FluA/FluB/RSV, and US IVD-marked Panther Fusion® SARS-CoV-2/Flu A/B/RSV Assays were evaluated using 405 nasopharyngeal UTM-preserved swabs. The LDRA demonstrated excellent concordance (overall agreement ≥ 98%, κ > 0.95), strong diagnostic accuracy, and reliable detection of mixed infections. This quadruplex provides a fully automated, rapid, and accurate solution for the simultaneous detection of influenza A, influenza B, SARS-CoV-2, and RSV viruses, enhancing molecular diagnostic capacity and supporting equitable, timely clinical decision-making in middle-income healthcare systems such as that of the Dominican Republic.

Abstract: Vision–language–action (VLA) models often suffer from limited robustness in long-horizon manipulation tasks due to their inability to explicitly exploit structural symmetries and to react adaptively when such symmetries are violated by environmental uncertainty. To address this limitation, this paper proposes PI-VLA, a symmetry-aware predictive and interactive VLA framework for robust robotic manipulation. PI-VLA is built upon three key symmetry-driven principles. First, a Cognitive–Motor Synergy (CMS) module jointly generates discrete and continuous action chunks together with predictive world-model features in a single forward pass, enforcing cross-modal action consistency as an implicit symmetry constraint across heterogeneous action representations. Second, a unified training objective integrates imitation learning, reinforcement learning, and state prediction, encouraging invariance to task-relevant transformations while enabling adaptive symmetry breaking when long-horizon deviations emerge. Third, an Active Uncertainty-Resolving Decider (AURD) explicitly monitors action-consensus discrepancies and state prediction errors as symmetry-breaking signals, dynamically adjusting the execution horizon through closed-loop replanning. Extensive experiments demonstrate that PI-VLA achieves state-of-the-art performance, attaining a 73.2% average success rate on the LIBERO benchmark and an 88.3% success rate in real-world manipulation tasks under visual distractions and unseen conditions. Ablation studies confirm that symmetry-aware action consensus and uncertainty-triggered replanning are critical to robust execution. These results establish PI-VLA as a principled framework that leverages symmetry preservation and controlled symmetry breaking to enable reliable and interactive robotic manipulation.

Abstract: Accurate assessment of emotional states is critical in clinical diagnostics, yet traditional multimodal sentiment analysis often suffers from "modality laziness," where models overlook subtle micro-expressions in favor of text priors. This study proposes Lingo-Aura, a cognitive-enhanced framework based on Mistral-7B designed to align visual micro-expressions and acoustic signals with large language model (LLM) embeddings. We introduce a robust Double-MLP Projector and global mean pooling to bridge the modality gap while suppressing temporal noise and ensuring numerical stability during mixed-precision training. Crucially, the framework leverages a teacher LLM to generate meta-cognitive label, such as reasoning mode and information stance, which are injected as explicit context to guide deep intent reasoning. Experimental results on the CMU-MOSEI dataset demonstrate that Lingo-Aura achieves a 135% improvement in emotion intensity correlation compared to text-only baselines. These findings suggest that Lingo-Aura effectively identifies discrepancies between verbal statements and internal emotional states, offering a powerful tool for mental health screening and pain assessment in non-verbal clinical populations.

Abstract: Background: Elevated low-density lipoprotein cholesterol (LDL-C) is a major risk factor for atherosclerosis and cardiovascular disease (CVD). Statins are the cornerstone of LDL-C reduction and are highly effective in secondary prevention. However, their benefit in primary prevention among individuals at low-to-moderate cardiovascular risk remains controversial, and long-term adherence is often limited by adverse effects. Methods: This narrative review summarizes current evidence on the clinical effectiveness of statin therapy, with particular attention to the role of nutritional status in modulating statin efficacy, safety, and interpretation of clinical outcomes. Results: In primary prevention among low- to moderate-risk populations, statin therapy often fails to demonstrate a clear reduction in cardiovascular events. Furthermore, 20–30% of patients in secondary or high-risk prevention do not achieve clinically meaningful benefits despite adequate LDL-C lowering. More than half of statin-treated patients discontinue therapy within two years, most commonly because of adverse effects, without a corresponding increase in cardiovascular mortality. Emerging evidence suggests that malnutrition and sarcopenia may significantly influence statin pharmacokinetics and pharmacodynamics, thereby affecting both therapeutic response and susceptibility to adverse events. In addition, statin-induced lipid lowering may alter nutrition-related biomarkers, potentially leading to misclassification or overestimation of malnutrition. Conclusions: Although statins remain effective agents for lowering LDL-C, their prescription should be embedded within an individualized, patient-centered approach. Current guidelines provide a robust methodological framework for statin use; however, their application should be contextualized rather than automatic. Optimal effectiveness is achieved when pharmacological therapy is integrated with dietary patterns, nutritional status, and lifestyle factors. Incorporating nutritional assessment into statin management may improve tolerability, enhance clinical outcomes, and enable more accurate cardiovascular risk stratification beyond standardized cholesterol-lowering strategies.

Abstract: Sepsis is a complex infection-driven inflammatory syndrome that can lead to life-threatening multi-organ failure and septic shock, characterized by cardiovascular instability, tissue hypoperfusion, and impaired oxygen utilization. Myocardial dys-function is a frequent and multifaceted complication in both adults and children, with sepsis-induced cardiomyopathy (SCM) representing an acute, reversible form of non-ischemic cardiac failure involving left and sometimes right ventricular impairment. Diagnosis relies on excluding acute coronary syndromes and recognizing refractory shock, low mixed venous oxygen saturation, and elevated cardiac biomarkers, espe-cially in patients with known risk factors such as pre-existing heart disease or elevated lactate. The pathophysiology reflects an interplay of systemic inflammation, circulatory redistribution, and mitochondrial dysfunction, while clinical recognition remains challenging, particularly in pediatrics where hypotension is a late sign. Early fluid re-suscitation is vital to restore perfusion, yet excessive administration risks fluid overload, underscoring the need for precise hemodynamic assessment. Conventional echocar-diographic measures of heart function may give misleading results in SCM because they depend on preload and afterload, often underestimating the true degree of myocardial impairment. Advanced hemodynamic monitoring, ranging from invasive to minimally invasive and non-invasive methods, is currently being studied for managing sepsis. Minimally invasive techniques offer detailed, dynamic data that complement echocar-diography and help identify specific hemodynamic profiles to guide septic shock treatment. Contemporary management increasingly favors multimodal monitoring and individualized strategies over protocol-driven approaches to optimize timely, goal-directed therapy.

Abstract: We present a formal analytical framework for the Riemann zeta function by mapping the Dirichlet η(s) function to a trace-class interaction operator Φ(s) on the Hilbert space l²(N). By applying a normalization kernel K(s), we establish a bijective mapping between the operator trace and the Riemann zeta function throughout the critical strip. We derive the Phase-Torque J(δ,t) representing the imaginary component of the interaction trace, and demonstrate that it vanishes identically on the critical line Re(s) = 1/2 due to unitary phase symmetry. Conversely, for Re(s) ≠ 1/2, a hyperbolic bias arises from the broken symmetry of the interaction magnitudes, which, when coupled with the Diophantine independence of prime logarithms, prevents the trace from vanishing. This geometric exclusion principle rigorously confines all non-trivial zeros to the critical line, providing a proof of the Riemann Hypothesis.

Abstract: Recent advances in reinforcement learning for large language models have produced powerful agent frameworks that achieve strong performance on multi-turn tool use, interactive search, and complex reasoning. However, existing reinforcement learning frameworks for large language model agents face three critical limitations: difficulty in handling dynamic user interactions owing to reliance on pre-scripted queries, limited scalability across varying interaction horizons with fixed scaling schedules, and substantial reward engineering overhead requiring domain-specific manual tuning. We introduce Progressive Multi-Turn Reinforcement Learning for Dynamic User-Interactive Tool Agents, a novel framework that integrates progressive user-interactive training to overcome sparse reward signals, adaptive horizon management that monitors performance metrics and adjusts training complexity accordingly, and domain-adaptive tool orchestration that learns optimal tool selection patterns across domains. Extensive experiments on WebArena, TAU-Bench, Berkeley Function-Calling Leaderboard Version 3, BabyAI, and SciWorld demonstrate that our method achieves 28.4% success rate on WebArena and 76.3\% on TAU-Bench, substantially outperforming the baselines, such as ReAct (16.2%) and MUA-RL (24.6%), while maintaining 94.7% performance on embodied reasoning tasks and 78.9\% cross-domain performance retention. Our work establishes a unified framework for realistic user interaction training, performance-adaptive complexity scaling, and domain-flexible tool orchestration.

Abstract: Trees in urban environments provide essential ecosystem services, but root growth–pavement system conflicts often constrain tree longevity and degrade infrastructure performance. The study presents a conceptual model for green and grey infrastructure alignment to ensure tree longevity while maintaining pavement performance in the urban environment. Drawing on past research where roots were flattened when exposed to confining stresses greater than 0.35 MPa, we developed a series of finite element models in COMSOL Multiphysics to simulate root-induced stresses in concrete pavements under varying pavement thickness, base thickness, and root depth. Parametric analyses showed that an increase in root depth had the largest impact in reducing stress, followed by an increase in pavement thickness, then base thickness. Maximum single-root-induced stresses were approximately 0.55 MPa, below that of normal concrete flexural strength. From these results, design guidance is proposed for tree root accommodation and pavement in existing and new infrastructure, with emphasis on root growth enhancement, pavement durability, and cost-effectiveness measures.

Abstract: Background/Objectives: Urology residency training widely varies across countries and evidence comparing residents’ experiences at an international level are limited. This study reports the results of an international survey of urology residents from different countries worldwide, aiming to characterize training environments, educational expo-sure, and trainee expectations across diverse healthcare systems. Methods: A 39-item online survey was administered to urology residents during the SIU Regional Meeting (Florence, November 2024), assessing demographics, training exposure, educational resources, workload, satisfaction, and career perspectives. Results were compared be-tween trainees at different postgraduate year (PGY) to explore associations for key out-comes. Results: Overall, 208 urology residents from 21 countries completed the survey. Most residents were actively involved in research (76.4%), although confidence in in-dependent scientific production was moderate (significantly lower among junior trainees). Surgical exposure increased with PGY, with good experience in endoscopy but limited hands-on exposure and expected autonomy in laparoscopic, robotic, and major open surgery. Despite high overall satisfaction with urology, residents described heavy workloads, inconsistent access to structured teaching and international fellowships, and a long-term shift in career expectations toward private practice. Conclusions: Urology residents worldwide report high engagement in research, strong satisfaction with their specialty choice, and interest in international mobility. Nonetheless, persistent disparities in surgical exposure, research confidence, workload, and gender representation highlight the need for competency-based curricula, structured mentorship, and improved training organization to promote equitable and high-quality urology education globally.

Abstract: Background/Objectives: Toothpaste ingredients such as strontium chloride (SrCl₂) and potassium carbonate (K₂CO₃) are recognized for their desensitizing and remineralizing ef-fects but may be absorbed through the oral mucosa. Their potential cytotoxic and cardio-toxic properties, however, remain inadequately characterized. Here, we investigated the effects of SrCl₂ and K₂CO₃ on mouse-induced pluripotent stem cells (iPSCs) and iPSC-derived cardiomyocytes (iPSC-CMs). Methods: Cells were exposed to varying con-centrations of each compound for up to 72 h. Real-time cell analysis (xCELLigence RTCA Cardio system) was used to assess proliferation, and flow cytometry was used to evaluate cell viability. Functional properties of iPSC-CMs were examined using multi-electrode ar-ray (MEA) recordings and the xCELLigence based impedance measurements. Cardiac marker expression was examined via immunofluorescence and quantitative RT-PCR. Results: Both SrCl₂ and K₂CO₃ affected iPSCs proliferation and reduced viability in a dose- and time-dependent manner, accompanied by altered embryoid body (EB) morphology and increased cell death. In iPSC-CMs, both compounds downregulated keys cardiac genes and disrupted spontaneous beating activity, with effects intensifying at the higher concentrations. Conclusions: These results demonstrate that SrCl₂ and K₂CO₃ induced dose-dependent cytotoxic and arrhythmogenic effects on iPSCs and iPSC-CMs. At elevated concentrations, these compounds impair iPSC-CMs function and may pose safety con-cerns upon chronic exposure. Further mechanistic and long-term in vivo studies are war-ranted to assess their potential cardiotoxic risk in consumer oral care products.

Abstract: The global rise of multidrug-resistant (MDR) bacterial infections in pediatric populations poses an alarming challenge to effective clinical management and antimicrobial stewardship. Two-component systems (TCSs), comprising sensor kinases and response regulators, play a pivotal role in bacterial adaptation, virulence, and resistance mechanisms, making them promising targets for diagnostic and therapeutic innovation. This study employs a machine learning-driven bioinformatics pipeline to identify and prioritize potential TCS biomarkers across MDR pediatric pathogens for integration into next-generation diagnostic biosensors. Genomic datasets from clinically relevant MDR bacteria were curated and analyzed to extract TCS-associated gene and protein signatures. Using Pfam domain features, multiple supervised learning models were trained, with XGBoost, Random Forest, and a Stacking Ensemble achieving high overall accuracies (0.9883-0.9885). While the dominant Non-TCS class was predicted with near-perfect accuracy, minority subclasses exhibited variable detection due to severe class imbalance, particularly for rare groups such as CpxA-like and EnvZ-like proteins (n=2 each). Moderate F1-scores were obtained for generic response regulators and OmpR-like proteins. Feature importance analysis identified a small set of highly discriminative domains, including PF01339, PF00702, PF07679, PF03997, and PF04886, associated with conserved regulatory and signaling motifs. These results demonstrate that Pfam domain signatures offer biologically meaningful features for TCS classification, while highlighting the need for expanded datasets or embedding-based features to improve minority-class prediction. Overall, this work provides a scalable, AI-driven foundation for TCS biomarker discovery, aiming to develop diagnostic biosensors for MDR pediatric pathogens.

Abstract: Public trust in the judiciary is fundamental for upholding the rule of law and ensuring democratic stability. However, in Puntland, Somalia, issues such as fairness, accessibility, and the influence of politics or clans may deter citizens from utilizing formal courts. This study assessed the level of public trust in the judiciary in Qardho, Garowe, and Bossaso. A cross-sectional mixed-methods approach was employed, involving a survey of 400 residents using a KOBO-based structured questionnaire and 12 key informant interviews with judges, lawyers, elders, and religious leaders. Quantita-tive data were analyzed through descriptive statistics, correlations, chi-square tests, and regression in Stata, while qual-itative data underwent thematic analysis. Overall, confidence was moderate: 62% agreed that the judiciary is fair and impartial, 55.25% had confidence in judges' independence, 63.5% trusted the enforcement of decisions, and 62.5% viewed processes as transparent. Confidence was most strongly linked to perceived enforcement (ρ = 0.730), judicial in-dependence (ρ = 0.699), and fairness (ρ = 0.686), with age (p = 0.001) and education (p < 0.001) significantly associated with confidence, unlike gender (p = 0.497) and work experience (p = 0.384). Enhancing decision enforcement, transpar-ency, access to information, and protections for judicial independence is vital for boosting public trust.

Abstract: We present a maximal referee-grade formulation of the Quantum Information Copy Time (QICT) program. All claims are restricted to (i) standard Axioms (locality, stationarity/KMS, conservation), (ii) executable certification predicates, or (iii) Theorems with fully enumerated premises. The key observable is the copy time \( \tau_{\text{copy}} \), defined operationally via Helstrom distinguishability. A certified hydrodynamic-window predicate CHW (constructed from finite-time witnesses and residual tests) gates every micro--macro statement. Under \( \)CHW we derive diffusion and prove the central scaling \( \tau_{\text{copy}} = \Theta\!\left(\sqrt{\chi^{(2)}_{\text{micro}}}\right) \), where \( \chi^{(2)}_{\text{micro}}=\langle \delta Q,\,(-\mathcal{L}_\perp)^{-2}\,\delta Q\rangle_{\mathrm{KM}} \) is a second-moment fast-complement susceptibility. Optional bridges (thermal modular saturation and Higgs-portal matching) are isolated as explicit model Axioms; the "Golden Relation'' is then a Theorem and is non-circular provided an independent \( \tau_{\text{copy}} \) inference NC is satisfied. We include an explicit, dataset-level certification appendix (tables generated from bundled validation outputs), enabling direct audit.

Abstract: Over the centuries, many birds have gone extinct, and many are currently endangered due to anthropogenic activities, inability of some birds to compete for food and the negative effects of climate change. To promote biodiversity of rare birds requires deliberate human efforts to create ecosystems that conserve them and enhance their survival. This work implemented a design-driven solution to an identified problem of squirrel feeding on bird seeds. Thus, it reports the design, development, prototyping and testing of a squirrel-proof birdfeeder capable of selectively preventing squirrels but allowing birds to feed from it. The design comprised of a compression spring and two concentric cylinders. Finite Element Analysis and Failure Mode and Effect Analysis were used to optimise the structural design and functionality of the bird feeder. Testing of the bird feeder showed that birds successfully fed from it, whilst squirrels could not access the feeds due to the mass differential mechanism based on Hooke’s law. Camera-recoded interactions showed that when a squirrel exerted its weight anywhere on the surface of the feeder, the spring compressed to displace the outside surface downwards to close-off the feeding holes and prevented a squirrel from accessing the bird seed. The prototype is a reliable solution to the problem of squirrels consuming bird seeds at home and in the parks.

Abstract: Mango holds a significant place globally. Considering its importance, it is also called 2 the king of fruits. Accurate price forecasting is essential for market decisions, policy 3 formulation, and agricultural market stability. Traditional time series models struggle 4 to provide effective and accurate forecasts of Mango prices and cannot capture their 5 nonlinear dynamics. The current study integrates machine learning, deep learning, and 6 statistical models to build a robust forecasting model using a dataset from 2001 to 2025. This 7 study proposed a novel attention-based Mango price forecasting approach. It significantly 8 forecasted Mango prices in the Indian market. It combines the strengths of various models 9 and produces generalized results. The hybrid ETS + ANN + GARCH model has high 10 predictive accuracy (MAE = 0.0498, MSE = 0.0106, RMSE = 0.1028, R2 209 = 0.774) and 11 ETS+SVM Hybrid achieves the accuracy level (MAE = 0.063, MSE = 0.006, RMSE = 0.078, 12 R2 = 0.873). The performance of ETS + BiLSTM is also significant, with an accuracy level 13 97.5%. Thus, an attention-based approach offers a new technological paradigm for mango 14 price forecasting.