Resource
2026 EN
Weipeng Jing · Wenjun Zhang · Xuancheng Zhang
+4 more
Effective foliage-wood separation plays a crucial role in forestry applications such as Leaf Area Index (LAI) estimation and Quantitative Structure Models (QSM). Point clouds provide valuable support for this task. However, large-scale scenes, uneven density, and occlusions hinder the use of general 3D vision methods. Existing Transformer-based methods typically partition point clouds into local patches, but the high computational complexity restricts the feasible patch size, often fragmenting tree structures and causing semantic information loss. Moreover, the geometric similarity of fine-scale foliage, coupled with limited context in small patches, makes feature discrimination more difficult. These two issues severely limit the performance of existing methods on foliage segmentation tasks. To address these challenges, we propose the Dynamic Hypergraph-guided Mamba (DHMamba) model with two key innovations. First, we leverage a lightweight Mamba-based architecture whose linear complexity enables processing of larger patches, thereby expanding the receptive field and reducing erroneous segmentation of branches and leaves. Second, we introduce a dynamic hypergraph-based serialization strategy to capture higher-order topological dependencies within local regions, enhancing the model's ability to extract discriminative features. Moreover, by introducing prior global density and designing two geometric feature descriptors, planarity and linearity, our framework further enhances the multi-scale discrimination of subtle differences in canopy. Extensive experiments on individual-tree and plot-scale datasets demonstrate that DHMamba substantially advances segmentation accuracy and robustness, highlighting its strong potential for practical large-scale forest point-cloud analysis and sustainable forest-resource management. Our code will be made available at https://github.com/ZhangIceNight/foliage_segmentation .
Resource
2026 EN
Ibrahim Alkhalifa · Razieh Eskandari · Mohamad Sawan
+1 more
Electrochemical biosensors are widely used for their small size, rapid response, and versatility, with techniques like chronoamperometry (CA) and voltammetry, at the forefront. A potentiostat applies an excitation waveform and measures the current flowing through electrochemical biosensors. Traditional potentiostat architectures consist of a control amplifier, a current sensing circuit, and an analog-to-digital converter (ADC). Recent advances have introduced new circuit designs and techniques that enhance performance and extend the range of applications. We highlight in this paper recent developments in single-mode potentiostat designs optimized for CA, cyclic voltammetry (CV), and fast-scan cyclic voltammetry (FSCV). We also explore the emerging class of digital potentiostats, discussing their advantages and limitations. Additionally, multimodal potentiostat designs, capable of simultaneously implementing multiple techniques such as CA, CV, FSCV, square wave voltammetry (SWV), and differential pulse voltammetry (DPV), are reviewed. The strengths, challenges, and potential applications of these systems are discussed. Finally, we present future research directions to address key challenges such as multi-channel high-throughput, drug delivery integration, CMOS/microfluidics packaging, and electrode biofouling, with the ultimate goal of enabling continuous in vivo monitoring.
Resource
2026 EN
Md Ibrahim Ibne Alam · Ankur Senapati · Anindo Mahmood
+2 more
Global Internet connectivity heavily relies on interconnection among Internet Service Providers (ISPs), achieved by accessing transit services or establishing direct peering relationships through Internet eXchange Points (IXPs). The latter offers more room for ISP-specific optimizations and is preferred but often involves a lengthy and convoluted process to set up peering agreements. Automating peering partner selection can greatly reduce the complexity. In this paper, we explore the use of publicly available data on ISPs to develop a machine learning (ML) approach that can predict whether ISP pairs should peer or not. First, we construct a large-scale dataset by processing and integrating information from public repositories (e.g., PeeringDB, CAIDA) and extract a diverse set of autonomous system-level features as inputs to ML models. We then evaluate the performance of three broad classes of ML models, i.e., tree-based, neural network-based, and transformer-based, to predict peering relationships. Among them, tree-based models achieve the best performance in our experiments, with XGBoost achieving a 98% accuracy, strong balanced-accuracy and F1-score performance in predicting peering partners. In addition, the model exhibits high robustness to variations in time, geographic region, and data incompleteness, indicating that it generalizes well in the rapidly evolving Internet landscape. We envision that ISPs can adopt our method towards automating their peering partner selection process, thus transitioning to a more efficient and optimized Internet ecosystem.
Resource
2026 EN
Amani Aldahiri · Veronika Stephanie · Ibrahim Khalil
+1 more
Facial recognition systems are widely used in device security, surveillance, and personalized services; however, they pose significant privacy risks. In resource-constrained Internet of Things (IoT) environments, limited local computing power often requires data to be transmitted for processing and model training, increasing vulnerability to attacks. This issue becomes critical when sensitive biometric data are transmitted and handled by third-party servers. Although Federated Learning (FL) enables collaborative model training without sharing raw data, it remains vulnerable to inference and reconstruction attacks and is often impractical for resource-limited devices. This paper presents SplitVFed, a framework that enhances privacy and cost-efficiency in collaborative learning for facial recognition. SplitVFed integrates Vertical Federated Learning (VFL) with Split Learning (SL) to distribute computation across clients, edge servers, and a central server. To reinforce privacy, Partial Secure Multi-Party Computation (P-SMPC) and lightweight Differential Privacy (DP) are employed to protect gradients and feature representations. The hybrid design mitigates gradient leakage while maintaining efficiency. Experimental results demonstrate that SplitVFed provides strong privacy preservation, reduced overhead, and competitive accuracy compared to state-of-the-art methods.
Journals
2026 EN
Huang Jianjie · Isah Kazeem O. · Olaniran Oladotun D.
+1 more
ABSTRACT To align with the global goal of keeping temperature rise well below 2°C, the European Emissions Trading System (EU‐ETS) was established as a market‐based initiative to mitigate climate change. While most carbon allowances are held and traded by polluting companies for compliance, financial actors including banks, investment firms, and brokers have increasingly participated as non‐emission compliance agents (ENCAs), engaging in speculative activities that may affect market functioning. Despite widespread discussion, compelling empirical evidence on the role of speculation in carbon price dynamics has been limited. Drawing from a large Google Trends dataset, we construct a news‐based speculation index to quantify speculative pressures and integrate it into an all‐inclusive modeling framework that accounts for both emission compliance and non‐compliance dynamics. Given the inherent volatility and mixed‐frequency nature of the data, we employ the GARCH‐MIDAS model, which accommodates variables at their natural frequencies while capturing both short‐ and long‐term volatility. Our out‐of‐sample forecasts show that models incorporating speculation and composite indicators consistently outperform the benchmark GARCH‐MIDAS‐RV model in returns, volatility, and Sharpe Ratios. Speculation emerges as a significant driver of risk‐adjusted performance, particularly over medium‐ and long‐term horizons, indicating that non‐compliance actors materially influence EUA price formation. The findings have clear economic and policy implications. Regulators can use data‐driven estimates of speculation to guide market participation rules and maintain stability, while investors can enhance portfolio performance by integrating speculative signals into forecasting models. Overall, the study provides a comprehensive, economically meaningful understanding of carbon price dynamics, reconciling market fundamentals with speculative activity and informing both policy and investment strategy in the EU‐ETS.
John Wiley & Sons Australia
Journals
2026 EN
Gu Yan · Chen Junfei · Xi Wenjia
+2 more
Abstract The relentless devastating impacts of global warming and other climate change effects leading to incessant ecological damages have compelled governments across the globe to rethink the pattern of natural resource depletion. This has motivated policymakers, governments, international organizations, and research pundits alike to advocate for a transition to sustainable consumption and production of natural resources. Consequently, there is a growing call for sustainable production and consumption practices, as outlined in Sustainable Development Goal 12 (SDG‐12). This research probes how natural resource production and consumption facilitate or hinder environmental sustainability in G7 economies from 1996 to 2020. The empirical evidence incorporates green energy, green technology, green finance, environmental tax, financial development, economic growth, and population as control variables within the STIRPAT theoretical framework. Second‐generation estimating techniques are utilized for empirical verification. An outstanding contribution of this study among others is the estimation of the moderating effects of green energy in mitigating the ensuing impact of natural resources on environmental sustainability. The results indicate that both production and consumption of natural resources, particularly coal and oil negatively affect environmental sustainability. Furthermore, green technology, energy, and finance as well as environmental tax are found to play a crucial role in promoting environmental sustainability. Green technology plays significant part in subduing the deteriorating effects of natural resources on the ecosystem. The robustness analyses further buttress the main analyses. Policy recommendations are proposed based on the empirical results.
Journals
2026 EN
Yilmaz Kars Merve · Kars Taha Ulutan · Guney Ibrahim
+3 more
ABSTRACT Aim This study investigated inflammatory markers involved in the pathogenesis of atherosclerosis in patients with Chronic Kidney Disease (CKD) undergoing maintenance hemodialysis (HD). Methods This cross‐sectional study included 144 patients with CKD (65 females, 79 males) undergoing maintenance HD and 23 healthy controls (12 females, 11 males). Serum concentrations of Transforming Growth Factor‐β (TGF‐β), C‐reactive protein (CRP), Neutrophil‐to‐Lymphocyte Ratio (NLR), and Platelet‐to‐Lymphocyte Ratio (PLR) were determined. Group comparisons were performed, and associations were examined using crude and multivariable‐adjusted logistic regression models. Results TGF‐β, NLR, and PLR levels were significantly higher in HD patients than in controls ( p < 0.001, p < 0.001, and p = 0.041, respectively), along with elevated CRP ( p < 0.001). ROC analysis showed good diagnostic performance for TGF‐β (AUC 0.812) and NLR (AUC 0.863), but weaker for PLR (AUC 0.633). In the crude logistic regression analysis, all three markers demonstrated significant associations with end‐stage renal disease (ESRD). After adjustment for age, sex, hypertension, cardiovascular disease (CVD), malignancy, and inflammatory diseases, TGF‐β remained significantly associated with ESRD (OR: 1.026; 95% CI: 1.003–1.050; p = 0.027). Similarly, after adjustment for age, sex, hypertension, CVD, diabetes mellitus (DM), malignancy, and inflammatory diseases, NLR was independently associated with ESRD (OR: 8.196; 95% CI: 1.554–43.227; p = 0.013). Finally, following adjustment for age, sex, hypertension, CVD, DM, and malignancy, PLR also showed a significant association with ESRD (OR: 1.019; 95% CI: 1.002–1.037; p = 0.033). Conclusion Inflammatory markers, including TGF‐β, NLR, and PLR, are significantly elevated in patients with CKD receiving maintenance HD compared with healthy individuals. These markers are associated with HD even after adjusting for significant covariates.
John Wiley & Sons Australia
Journals
2026 EN
Tuzcu Tuba Uğur · Karaduman İbrahim · Kardaş Rıza Can
+4 more
ABSTRACT Objectives Familial clustering and HLA haplotype association studies suggest that genetic factors disrupting the regulation of the immune system may predispose to risk of both neoplasms and autoimmune diseases, potentially leading to an increased frequency of cancer development in patients with primary Sjögren's disease (SjD), as well as their relatives. In this study, we aimed to assess the risk of cancer in patients with primary SjD and their close relatives. Methods Primary SjD patients who were actively followed‐up in the rheumatology outpatient clinic at Gazi University Hospital and who met the 2016 ACR‐EULAR classification criteria for primary SjD were included in the study. Data on cancer history in patients and their relatives were collected through direct face‐to‐face interviews and telephone surveys with the patients. The risk of developing cancer was calculated by comparing it with the general population of Turkey obtained from the Global Cancer Observatory of the World Health Organization International Agency for Research on Cancer (GLOBOCAN). Results A total of 323 primary SjD patients (F/M: 313/10, mean age: 56 ± 11) and their 1750 close relatives (parents, siblings, and children) were studied. Among SjD patients, 29 (9%) had a history of malignancy. Of these, 19 (5.9%) were solid organ and 10 (3.1%) were hematological malignancies. Breast cancer was the most common solid tumor. The median follow‐up was 3.6 years, and the calculated standardized incidence ratio (SIR) for all cancers was 3.3 (95% CI: 2.2–4.7, p < 0.001). Leukemia or lymphoma cases had an SIR of 22.5 (95% CI: 10.8–41.4, p < 0.001). Among 313 women, seven cases of breast cancer had an SIR of 3.8 (95% CI: 1.5–7.9, p < 0.001). Risk of malignancy in patients with SjD did not differ based on age, gender, smoking history, Schirmer test result, laboratory parameters including anti‐SSA, anti‐SSB, ANA, complement levels, ESSDAI status, or Focus score, but it was higher in the presence of a cancer among close relatives. A total of 128 (43.3%) patients with SjD had at least one close relative with cancer (176 cancer cases in total), giving an SIR of 3.5 (95% CI: 3.0–4.1, p < 0.001). The average age of close relatives with cancer was 58 ± 10 years; 56% were male, and 7.4% were active smokers. Conclusion Our results suggest that not only patients with primary SjD but also their close relatives have an increased risk of developing cancer compared with the general population.
Journals
2026 EN
Ibrahim Rita · Froschauer Christin · Broschk Susanne
+2 more
ABSTRACT The changing demography of human populations has motivated a search for interventions that promote healthy ageing, and especially for evolutionarily‐conserved mechanisms that can be studied in lab systems to generate hypotheses about function in humans. Reduced Insulin/IGF signalling (IIS) is a leading example, which can extend healthy lifespan in a range of animals, but whether benefits and costs of reduced IIS vary genetically within species is under‐studied. This information is critical for any putative translation. Here, in Drosophila , we test for genetic variation in lifespan response to a dominant‐negative form of the insulin receptor, along with a metric of fecundity to evaluate corollary fitness costs/benefits. We also partition genetic variation between DNA variants in the nucleus (nDNA) and mitochondrial DNA (mtDNA), in a fully‐factorial design that allows us to assess ‘mito‐nuclear’ epistasis. We show that reduced IIS can have either beneficial or detrimental effects on lifespan, depending on the combination of mtDNA and nDNA. This suggests that, while insulin signalling has a conserved effect on ageing among species, intraspecific effects can vary genetically, and the combination of mtDNA and nDNA can act as a gatekeeper.
Journals
2026 EN
Shikuku Kelvin Mashisia · Ochenje Ibrahim · Osiemo Jamleck
+3 more
ABSTRACT Considerable attention has been placed on bundled index insurance to enhance climate resilience, address multiple risks simultaneously, and increase the adoption of agricultural technologies. We conducted an endow‐and‐exchange choice experiment with 1,828 female and male livestock keepers in northern Kenya to elicit their preferences for bundled index‐based livestock insurance (IBLI). We measured relative willingness to pay (WTP) as the maximum amount of money that an individual is willing to pay to switch from one bundle to another. We found that livestock keepers were willing to pay 19%–33%, 100%–153%, and 148%–232% more for IBLI + animal nutrition , IBLI + animal health , and IBLI + flexible package , respectively, relative to IBLI + animal breed . Relative to the average WTP to switch from other bundles to IBLI + animal breed , women had 36%–45%, 54%–64%, and 76%–84% higher WTP than men for IBLI + animal nutrition , IBLI + animal health , and IBLI + flexible package , respectively. Providing information about bundled products and seasonal vegetation forecasts reduced the relative WTP for IBLI + animal nutrition . Our findings highlight the importance of considering the differential preferences of women and men when designing and promoting bundled IBLI products.