Search Results (412)

This study addresses the issues of sparse observations from buoys in the tropical Indian Ocean and systematic biases in reanalysis products by proposing a daily-mean wind speed reconstruction framework that integrates multi-source meteorological fields. This study also considers the impact of different source domains on model pre-training, with the goal of providing reliable data support for wind energy assessment. The model was pre-trained using data from the Americas and tropical Pacific buoys as the source domain and then fine-tuned on Indian Ocean buoys as the target domain. Using annual leave-one-out cross-validation, we evaluated the model’s performance against uncorrected ERA5 and CCMP data while comparing three deep reconstruction models. The results demonstrate that deep models significantly reduce reanalysis bias: the RMSE decreases from approximately 1.00 m/s to 0.88 m/s, while R² improves by approximately 8.9% and 7.1% compared to ERA5/CCMP, respectively. The Branch CNN–Transformer outperforms standalone LSTM or CNN models in overall accuracy and interpretability, with transfer learning yielding directional gains for specific wind conditions in complex topography and monsoon zones. The 20-year wind energy data reconstructed using this model indicates wind energy densities 60–150 W/m² higher than in the reanalysis data in open high-wind zones such as the southern Arabian Sea and the Somali coast. This study not only provides a pathway for constructing high-precision wind speed databases for tropical Indian Ocean wind resource assessment but also offers precise quantitative support for delineating priority development zones for offshore wind farms and mitigating near-shore engineering risks. Full article

As corporate sustainability reporting evolves into a pivotal resource for investors, regulators, and stakeholders, the imperative to evaluate and elevate ESG disclosure quality intensifies amid persistent challenges like opacity, inconsistency, and greenwashing. This review synthesizes interdisciplinary insights from accounting, finance, and computational linguistics on artificial intelligence (AI), particularly natural language processing (NLP) and machine learning (ML), as a transformative force in this domain. We delineate ESG disclosure quality across four operational dimensions: readability, comparability, informativeness, and credibility. By integrating cutting-edge methodological innovations (e.g., transformer-based models for semantic analysis), empirical linkages between AI-extracted signals and market/governance outcomes, and normative discussions on AI’s auditing potential, we demonstrate AI’s efficacy in scaling measurement, harmonizing heterogeneous narratives, and prototyping greenwashing detection. Nonetheless, causal evidence linking managerial AI adoption to stakeholder-perceived enhancements remains limited, compounded by biases in multilingual applications and interpretability deficits. We propose a forward-looking agenda, prioritizing cross-lingual benchmarking, curated greenwashing datasets, AI-assurance pilots, and interpretability standards, to harness AI for substantive, equitable improvements in ESG reporting and accountability. Full article

Accurate interpretation of multimodal wearable data remains challenging in occupational environments due to heterogeneous sensing modalities, motion artifacts, and dynamic work conditions. This study proposes and validates an adaptive multimodal clustering framework for occupational health monitoring. The framework jointly models physiological, activity, and location data from 24 highway-maintenance workers, incorporating a silhouette-guided feature-weighting mechanism, multi-scale temporal change-point detection, and KDE-based spatial analysis. Specifically, the analysis identified three distinct and interpretable behavioral–physiological states that exhibit significant physiological differences (p < 0.001). Notably, it reveals a predominant yet heterogeneous baseline state alongside acute high-intensity and episodic surge states, offering a nuanced view of occupational risk beyond single-modality thresholds. The integrated framework provides a principled analytical workflow for spatiotemporal health risk assessment in field settings, particularly for vibration-intensive work scenarios, while highlighting the complementary role of physiological indicators in low- or static-motion tasks. This framework is particularly effective for vibration-intensive tasks involving powered tools. However, to mitigate potential biases in detecting static heavy-load activities with limited wrist motion (e.g., lifting or carrying), future extensions should incorporate complementary weighting of physiological indicators such as heart rate variability. Full article

Visual large language models (VLLMs) are discussed as potential tools for assisting radiologists in image interpretation, yet their clinical value remains unclear. This study provides a systematic and comprehensive comparison of general-purpose and biomedical VLLMs in radiology. We evaluated 180 representative clinical images with validated reference diagnoses (radiography, CT, MRI; 60 each) using seven VLLMs (ChatGPT-4o, Gemini 2.0, Claude Sonnet 3.7, Perplexity AI, Google Vision AI, LLaVA-1.6, LLaVA-Med-v1.5). Each model interpreted the image without and with clinical context. Mixed-effects logistic regression models assessed the influence of model, modality, and context on diagnostic performance and hallucinations (fabricated findings or misidentifications). Diagnostic accuracy varied significantly across all dimensions (p ≤ 0.001), ranging from 8.1% to 29.2% across models, with Gemini 2.0 performing best and LLaVA performing weakest. CT achieved the best overall accuracy (20.7%), followed by radiography (17.3%) and MRI (13.9%). Clinical context improved accuracy from 10.6% to 24.0% (p < 0.001) but shifted the model to rely more on textual information. Hallucinations were frequent (74.4% overall) and model-dependent (51.7–82.8% across models; p ≤ 0.004). Current VLLMs remain diagnostically unreliable, heavily context-biased, and prone to generating false findings, which limits their clinical suitability. Domain-specific training and rigorous validation are required before clinical integration can be considered. Full article

(1) Background: The use of wearable technology for assessing running biomechanics in field-based sports has increased in recent years. Inertial measurement units (IMUs) are low-cost, non-invasive devices capable of estimating spatiotemporal gait-related metrics during overground locomotion. This study evaluated the accuracy and concurrent validity of a foot-mounted IMU system for estimating sprinting kinematics. (2) Method: Twenty-five elite and sub-elite athletes completed four maximal 10-metre fly efforts, with their kinematics measured concurrently using a three-dimensional motion analysis system and IMUs. (3) Result: The foot-mounted IMU system’s root mean square errors for stride length and duration were 0.22 m and 0.04 s, respectively. Mean biases (95% level of agreement) were −0.67 $\mathrm{m} · {\mathrm{s}}^{-1}$ (−1.19; −0.14) for peak velocity, −0.51 $\mathrm{m} · {\mathrm{s}}^{-1}$ (−1.10; 0.09) for instantaneous velocity, and 0.17 $\mathrm{m} · {\mathrm{s}}^{-2}$ (−1.04; 1.37) for instantaneous acceleration. Stride length, duration, and cadence were −0.07 m (−0.36; 0.23), 0.02 s (−0.02; 0.06), and −4.64 strides $· {\min }^{-1}$ (−15.82; 6.53), respectively. (4) Conclusions: End users implementing this technology in research and practice should interpret this study’s findings relative to their analytical objectives, logistical resources, and operational constraints. Therefore, its adoption should be guided by the specific performance metrics of interest and the extent to which the system’s capabilities align with the outcomes the end user aims to achieve. Full article

Background/Objectives: Early larval development is critical for caste and sex differentiation in honeybees. This study investigates molecular divergence in 4-day-old Apis mellifera larvae and introduces a customized deep learning model for hub-gene discovery. Methods: Genome-guided RNA-seq, DEGs, WGCNA, and splicing analyses were integrated. A hybrid convolution–attention model, ACmix-Swin, combined with WGAN-GP augmentation, was developed to classify larvae and prioritize caste-biased genes. Selected genes were validated by qPCR. Results: Significant caste- and sex-specific divergence was detected in cuticle formation, hormone metabolism, and reproductive signaling. ACmix-Swin achieved the highest accuracy among baseline models and consistently identified key regulators, including Vg, LOC725841, LOC412768, and LOC100576841. qPCR confirmed RNA-seq trends. Conclusions: Caste- and sex-specific transcriptional programs are established early in larval development. The ACmix-Swin framework provides an effective strategy for high-dimensional transcriptome interpretation and robust hub-gene identification. Full article

Background/Objectives: Marker-based motion capture remains widely used for lower limb kinematics due to its high precision, although its application is often constrained by elevated operational costs and the requirement for controlled laboratory environments. Markerless methods, such as MediaPipe offer a promising alternative for extending biomechanical analyses beyond traditional laboratory settings, but evidence supporting their validity in controlled tasks is still limited. This study aimed to validate a pixel-based markerless pipeline for two-dimensional kinematic analysis of hip and knee motion during squatting. Methods: Ten healthy volunteers performed three squats with a maximum depth of 90°. Kinematic data were collected simultaneously using marker-based and markerless systems. For the marker-based method, hip and knee joint angles were calculated from marker trajectories within a fixed coordinate system. For the markerless approach, a custom pixel-based pipeline was developed in MediaPipe 0.10.26 to compute bidimensional joint angles from screen coordinates. A paired t-test was conducted using Statistical Parametric Mapping, and maximum flexion values were compared between systems with Bland–Altman analysis. Total range of motion was also analyzed. Results: The markerless pipeline provided valid estimates of hip and knee motion, despite a systematic tendency to overestimate joint angles compared to the marker-based system, with a mean bias of −17.49° for the right hip (95% LoA: −51.89° to 16.91°). Conclusions: These findings support the use of markerless tools in clinical contexts where cost and accessibility are priorities, provided that systematic biases are taken into account during interpretation. Overall, despite the systematic differences, the 2D MediaPipe-based markerless system demonstrated sufficient consistency to assist clinical decision-making in settings where traditional motion capture is not available. Full article

We present an integrated methodology for three-dimensional inversion of large-scale airborne electromagnetic (AEM) and magnetic survey data that simultaneously recovers electrical conductivity, chargeability, and both induced and remanent magnetizations. A central feature of the AEM component is the explicit incorporation of induced polarization (IP) effects. Neglecting IP responses can lead to biased conductivity models, particularly in mineralized systems where disseminated sulfides contribute strongly to chargeability. Using the Generalized Effective-Medium Theory of Induced Polarization (GEMTIP), the inversion produces physically consistent 3D distributions of conductivity and chargeability. To enhance magnetic interpretation, we also implement a vector magnetic inversion that resolves both induced and remanent magnetization from Total Magnetic Intensity (TMI) data, enabling geologically realistic magnetization models in terranes with significant remanence. This integrated workflow was applied to airborne AEM and TMI datasets collected over the Asankrangwa Gold Belt in central Ghana. The inversion results delineate a key exploration target defined by coincident magnetic low and elevated chargeability, interpreted as sulfide-rich gold mineralization and subsequently confirmed by drilling. These results demonstrate that jointly accounting for IP and remanent magnetization in 3D inversion substantially improves subsurface characterization and provides a powerful tool for mineral exploration in structurally and lithologically complex environments. Full article

Background/Objectives: Optimal selection for perioperative therapy in urothelial carcinoma (UC) remains uncertain. We evaluated the efficacy of neoadjuvant and/or adjuvant chemotherapy (NAC/AC) for patients with bladder cancer (BC) and upper tract UC (UTUC), examined the role of lymphovascular invasion (LVI), and considered the implications for adjuvant nivolumab. Methods: We retrospectively analyzed consecutive patients who underwent radical cystectomy or radical nephroureterectomy at a single center (July 1998–April 2021; observation to 31 March 2025). After exclusions, 252 BC and 153 UTUC patients were included. Endpoints were cancer-specific survival, progression-free survival (PFS; BC), non-urinary-tract recurrence-free survival (NUTRFS; UTUC), and overall survival (OS). Survival was estimated by Kaplan–Meier analysis and compared by log-rank tests. Results: For BC, AC did not improve the PFS or OS in the overall pT ≥ 2 population, whereas node-positive (pN+) disease derived significant benefits in both endpoints among NAC-naïve patients (PFS and OS, p = 0.002 and p = 0.008). For UTUC, AC conferred no advantage in NUTRFS or OS for the overall pT ≥ 2 population. However, NUTRFS benefits emerged in the pN+ subset (p = 0.049), although the OS was not improved. Among NAC-treated BC, the outcomes were poorest for ≥ypT3 and ypN+, whereas ypT ≤ 2 fared better. LVI was associated with adverse outcomes and was borderline higher in pN+ versus pT ≥ 2/pN− for BC (p = 0.056) and significantly higher for UTUC (p = 0.012). Conclusions: In this retrospective, single-center cohort, our exploratory analyses suggest that perioperative benefit is largely node-dependent, supporting prioritizing systemic therapy for pN+ disease and cautioning against routine AC for pT2/ypT2 without nodal involvement. After NAC, adjuvant therapy appeared most justified for ≥ypT3/ypN+. Prospective biomarker-integrated validation is warranted and, given the small and underpowered subgroups and the potential for selection and immortal time biases, these observations should be interpreted as hypothesis-generating rather than causal. Full article

Background/Objectives: The so-called ‘Western diet’ characterised by the frequent consumption of high energy-dense (HED) food is linked with overeating, obesity, and an array of physiological and weight-related health complications. Attentional biases to HED food, which have been identified as a key mechanism promoting overeating, arise when reward-driven automatic processes impair the internal states responsible for regulating hunger and satiety. Emerging mindfulness-based interventions show promise in attenuating attentional biases by training controlled processes and enhancing the self-regulatory mechanisms required to override reward-driven automatic processing. Methods: Following PRISMA 2020 guidelines and PICOS strategy, this systematic review collates and synthesises current research on the impact of mindfulness interventions on visual attention and attentional bias to food cues in adults. Searches were conducted in Web of Science, PubMed, Scopus, Springer Nature, MEDLINE, Embase, and CINAHL in September 2025. Results: Findings obtained from six eligible studies were mixed indicating that mindfulness interventions significantly reduced attentional bias to HED, whereas other interventions indirectly enhanced self-regulatory systems such as hedonic hunger and craving without directly modifying attention. Additional findings highlight reductions in physiological reactivity, increased interoceptive awareness, and savouring. Conclusions: Overall findings suggest that mindfulness-based practices hold preliminary but promising potential to subdue attentional biases to HED food and disrupt unhealthy eating habits influenced by the Western diet. However, the limited number of studies, small sample sizes, methodological heterogeneity, and lack of mechanistic clarity indicate that such conclusions should be interpreted with caution. More robust and standardised research is warranted to determine whether mindfulness can produce durable, real-world behavioural change. Full article

This paper asks whether debt among affluent consumers reflects rational leverage, comparable to firms, or the influence of cognitive biases. Using survey data on Brazilian bank clients, we combine logistic regressions with a finite-mixture-inspired, rule-based classification and a test based on a ten-business-day overdraft grace period to identify heterogeneity in borrowing behavior. In the high-income subsample, Cognitive Reflection Test scores are unrelated to debt incidence, diverging from prior evidence in mixed-income populations. Among indebted affluent respondents, most borrowing is cost-sensitive and consistent with deliberate leverage (about 80 percent), while a minority displays patterns consistent with optimism bias and overconfidence (about 20 percent). The institutional feature of a temporary grace period lowers the effective cost of short-term credit and is associated with a marked reduction in overdraft use, reinforcing the leverage interpretation. Overall, consumer debt is heterogeneous; for the affluent, it largely aligns with leverage, though behavioral biases persist at the margins. Policy for high-income borrowers should prioritize targeted measures that address optimism bias and overconfidence while preserving deliberate leverage management through clear disclosures and monitoring of sensitivity to short-term credit costs. Full article

Quantitative Polymerase Chain Reaction (qPCR) is a very sensitive method to determine small amounts of DNA or RNA in experimental, environmental, veterinary, forensic and clinical samples. Despite efforts from the qPCR community to address qPCR variability by recommending standardization of reporting of all steps of a qPCR experiment, most reported qPCR results are still grossly biased. The first part of this paper describes two decades of efforts to remedy this situation by promoting so-called efficiency-corrected qPCR data analysis. Although such analysis leads to less variable qPCR results, the outcome, fluorescence at cycle zero, is difficult to grasp. In the second part, we outline how qPCR analyses can result in N_copy, the number of copies of the target at the start of the reaction. A newly developed theoretical approach determines N_copy using the characteristics of the amplification curve and the known concentrations of all reaction components. By including these reaction-mix characteristics in the analysis, this N_copy is assay-, machine- and laboratory-independent and thus allows direct worldwide comparisons. Moreover, N_copy provides a very intuitive and easy-to-interpret absolute quantitative result. Full article

BACKGROUND. Managing women in pregnancy or labour is becoming a serious challenge because of delayed conception age and higher morbidity. The main negative factor is increasing numbers of overweight and obese women. Fatty tissue significantly biases the detection of uterine contractions by tocography, which is routinely used in obstetrical wards. Thus, the FDA approved an alternative method called electrohysterography (EHG) and recommended it for women with an over-normal BMI. However, almost all published methods of labour prediction based on EHG signals use a “black-box model” approach, i.e., increasingly numerically complex signal features and classification algorithms that are chosen a priori, without any physiological rationale behind them. This makes using these algorithms difficult in obstetrical practice. AIM. The aim of the study was to show that a simple classifier based on a single and physiologically interpretable parameter can predict uterine contractions during labour with an accuracy comparable to advanced classifiers. METHODS. An obstetrical interpretable EHG parameter was introduced called the uterine activity index. To avoid the influence of confounding factors associated with preterm labour and imbalanced signal sets, this classifier was evaluated using the private, retrospective database of EHG signals registered for 45 women in the third trimester of a pregnancy, and 31 women in the second stage of labour with a normal BMI. The classifier, based on the logistic regression model, was tested using the bootstrap method. RESULTS. The bootstrapping mean (95% confidence interval) of the AUC ROC estimated for the 200 bootstrap samples was 0.96 (0.91–0.99). This accuracy was slightly better for EHG signals in comparison to predictions based on classical tocography. CONCLUSIONS. The obtained results confirm that a simple physiologically explained classifier can be considered in commercial applications of electrohysterography. However, its clinical significance should be evaluated through properly designed randomised clinical trials. Full article

This review investigates the application of Explainable Artificial Intelligence (XAI) in biomedical informatics, encompassing domains such as medical imaging, genomics, and electronic health records. Through a systematic analysis of 43 peer-reviewed articles, we examine current trends, as well as the strengths and limitations of methodologies currently used in real-world healthcare settings. Our findings highlight a growing interest in XAI, particularly in medical imaging, yet reveal persistent challenges in clinical adoption, including issues of trust, interpretability, and integration into decision-making workflows. We identify critical gaps in existing approaches and underscore the need for more robust, human-centred, and intrinsically interpretable models, with only 44% of the papers studied proposing human-centred validations. Furthermore, we argue that fairness and accountability, which are key to the acceptance of AI in clinical practice, can be supported by the use of post hoc tools for identifying potential biases but ultimately require the implementation of complementary fairness-aware or causal approaches alongside evaluation frameworks that prioritise clinical relevance and user trust. This review provides a foundation for advancing XAI research on the development of more transparent, equitable, and clinically meaningful AI systems for use in healthcare. Full article

Plant tissues exhibit a layered architecture that makes spatial context decisive for interpreting transcriptional changes. This review explains why the location of gene expression is as important as its magnitude and synthesizes advances uniting single-cell/nucleus RNA-seq with spatial transcriptomics in plants. Surveyed topics include platform selection and material preparation; plant-specific sample processing and quality control; integration with epigenomic assays such as single-nucleus Assay for Transposase-Accessible Chromatin using sequencing (ATAC) and Multiome; and computational workflows for label transfer, deconvolution, spatial embedding, and neighborhood-aware cell–cell communication. Protoplast-based single-cell RNA sequencing (scRNA-seq) enables high-resolution profiling but introduces dissociation artifacts and cell-type biases, whereas ingle-nucleus RNA sequencing (snRNA-seq) improves the representation of recalcitrant lineages and reduces stress signatures while remaining compatible with multiomics profiling. Practical guidance is provided for mitigating ambient RNA, interpreting organellar and intronic metrics, identifying doublets, and harmonizing batches across chemistries and studies. Spatial platforms (Visium HD, Stereo-seq, bead arrays) and targeted imaging (Single-molecule fluorescence in situ hybridization (smFISH), Hairpin-chain-reaction FISH (HCR-FISH), Multiplexed Error-Robust Fluorescence In Situ Hybridization (MERFISH)) are contrasted with plant-specific adaptations and integration pipelines that anchor dissociated profiles in anatomical coordinates. Recent atlases in Arabidopsis, soybean, and maize illustrate how cell identities, chromatin accessibility, and spatial niches reveal developmental trajectories and stress responses jointly. A roadmap is outlined for moving from atlases to interventions by deriving gene regulatory networks, prioritizing cis-regulatory targets, and validating perturbations with spatial readouts in crops. Together, these principles support a transition from descriptive maps to mechanism-informed, low-pleiotropy engineering of agronomic traits. Full article