Search Results (1,777)

Forest landscapes used for recreation also provide climate-regulating services that are often absent from spatial planning. We mapped aboveground biomass density (AGBD) and aboveground carbon storage in the Postavaru-Piatra Mare mountain landscape near Brașov, Romania, using a cross-platform remote-sensing workflow. MLS point clouds from 16 plots served as an independent plot-level reference, UAV LiDAR acquired at 70, 120, and 150 m provided airborne comparisons, and GEDI L4A samples, Sentinel-2 predictors, and terrain variables were modelled in Google Earth Engine using Random Forest (RF) and Gradient Boosted Trees (GBT). GBT achieved R² = 0.34 and RMSE = 106.42 Mg ha⁻¹, whereas RF achieved R² = 0.33 and RMSE = 107.45 Mg ha⁻¹. These values indicate limited explanatory power; accordingly, the resulting 25 m maps are interpreted as landscape-scale screening products rather than operational stand inventories. Against MLS, RF produced an RMSE of 79.05 Mg ha⁻¹ and GBT 82.58 Mg ha⁻¹, while UAV estimates were positively biased at all tested flight heights. Mean carbon density peaked at 1200–1400 m and was highest in coniferous forest, whereas broadleaved and tree-cover classes contributed the largest total stock because of their larger area. The RF and GBT estimates differed by 2.3% in aggregated study-area carbon stock. Aboveground carbon storage was the only ecosystem service quantified. The resulting maps should be interpreted as landscape-scale screening layers and as ecological inputs for future studies integrating trail, visitor-use, participatory-mapping or perception data; they do not quantify cultural ecosystem services. Full article

Ordered-risk assessment is a safety-critical learning problem in which delayed or biased risk estimation can affect prioritization and response planning. Existing expert-knowledge and data-driven approaches face a trade-off among transparent indicator design, nonlinear representation capability, and stable optimization when reliable labeled data are scarce. To address this gap, we developed a two-stage adaptive differential evolution-optimized extreme learning machine framework (TA-DE-ELM) for six-level ordered-risk assessment and evaluated it in a controlled physics-informed synthetic simulation benchmark. The benchmark encodes kinematic, capability, sensing/interference, and resilience priors as explicit scoring rules for model evaluation, rather than as an application simulator. The method combines transparent risk-logic specification, a stage-wise exploration–refinement optimizer with stagnation-triggered restart, and a validation objective that jointly considers cross-entropy, macro-F1, ordinal error, and accuracy. Under a unified finite budget, TA-DE-ELM ranked first among all tested ELM-family baselines for accuracy, macro-F1, quadratic weighted kappa, and ordinal mean absolute error, with paired tests indicating improvements over the closest competitor ($p<0.013$). These results show that TA-DE-ELM can recover an expert-rule-induced ordered-risk mapping more effectively than the tested baselines under controlled finite-sample conditions. Further validation with higher-fidelity simulators, externally collected datasets, and richer temporal perturbation protocols remains necessary before application-specific use. Full article

Stereo vision measurement in industrial robot environments fails on thin structures such as needles and gripper tips, because both traditional and deep-learning-based stereo matching produce boundary errors there. We identify and verify experimentally that ground-truth disparity inflation in public datasets is a systematic error source that pushes networks toward biased boundary estimates. From this, we classify the resulting measurement deviations into four categories: edge errors, intra-segment expansion, small-object loss, and uniform-region distortion. We then propose a geometric-constraint-driven measurement refinement method that corrects each category in turn. Since every correction step is geometric rather than learned, the method is unaffected by ground-truth inflation, a property that conventional post-processing filters do not offer. A GUM-based uncertainty propagation analysis of the measurement model $D=f·B/d$ shows that disparity uncertainty dominates the depth uncertainty budget when f and B are exactly known. Experiments on KITTI 2015, Middlebury, and a custom UE4 synthetic industrial dataset (100 stereo pairs) with nine stereo baselines (seven deep learning and two traditional) show that, on inflation-free ground truth, the refinement imposes a near-zero systematic penalty on deep learning output while clearly improving traditional methods. On KITTI, the predictable metric shift confirms that the method is unaffected by LiDAR ground-truth inflation. On a real industrial robot scene, the refined disparity recovers the gripper tip and needle that the baseline LEAStereo loses. These results position geometric-constraint-driven refinement as an effective, training-data-independent complement to end-to-end stereo matching for precision industrial measurement, within the tested scenes and methods. Full article

Accurate correction of daily satellite-derived precipitation estimates in data-scarce tropical regions remains a critical challenge for climate monitoring, agriculture, and public health. Satellite products such as CHIRPS offer broad spatial coverage but exhibit systematic biases relative to ground-based observations particularly in complex terrain under bimodal tropical regimes influenced by ENSO. We propose a Functional Generalised Additive Mixed Model (FGAMM) that corrects CHIRPS-derived precipitation estimates by treating the annual accumulated precipitation curve as a functional response and the satellite accumulation curve as a functional covariate, while incorporating station-level random effects and the Southern Oscillation Index. This functional formulation targets the systematic, slowly varying bias between satellite and ground-station accumulation, the quantity most relevant for water-balance applications such as reservoir management and agricultural planning rather than day-to-day storm nowcasting. Applied to 62 IDEAM stations in the Valle del Cauca department of Colombia (2012–2020), the FGAMM achieves a mean cross-validation RMSE of 0.68 mm/day (95% bootstrap CI: 0.61–0.75), a substantially lower error than linear regression, SVM, and Random Forest within this dataset, where the gap is statistically significant across all competing methods. This magnitude of advantage is not reproduced when applying the same fitting-and-differencing pipeline, via a simplified concurrent approximation, to an independent national-network dataset; we discuss the methodological factors that likely contribute to this discrepancy—including an inherent smoothness asymmetry between the penalised-spline FGAMM fit and the unconstrained benchmark models, and differences in validation design between the two checks—in the Discussion, and treat the true size of the FGAMM’s advantage as an open question pending a fully controlled comparison. Corrected estimates are currently restricted to the calibrated station locations; because CHIRPS provides near-global daily coverage from 1981 to the present, we discuss how the same modelling approach could in principle be applied to other tropical or subtropical regions with a sparse reference station network, including areas of Latin America, sub-Saharan Africa, and South Asia where station density is similarly limited. Full article

Musculoskeletal injuries are the leading cause of time loss in professional football, yet machine learning models integrating multi-protocol force-plate data for injury-risk classification are absent from the Latin American professional football literature. Purpose: This study aims to develop and internally cross-validate a supervised Random Forest model for injury risk classification using full-season VALD ForceDecks data from a professional Costa Rican club, with thresholds derived from receiver operating characteristic (ROC) analyses. Methods: We utilized an observational longitudinal study (September 2024 to April 2026). One hundred and twenty-six male footballers from five competitive categories underwent 518 dual-force-plate assessments across four protocols (Nordic Hamstring, isometric mid-thigh pull, isometric adductor squeeze, countermovement jump), yielding 21 variables; one assessment per player entered the model. Record linkage against 262 surveillance-documented time-loss events identified 53 players with pre-injury assessments as the positive class (42.1%; ratio 1:1.4). SMOTE (k = 5) was applied within each training fold only. A Random Forest model (500 trees; depth 4) used 5-fold stratified cross-validation. The HIGH boundary was set at the Youden index, and the LOW boundary was set as the first score quartile. Results: The mean AUC-ROC across folds was 0.683 +/− 0.050, the only estimate of generalisation reported. Refitted on the complete dataset and applied back to the same players, the model gave at the Youden cut-point (score ≥ 45.2) an apparent sensitivity of 100% and specificity of 95.9%; these resubstitution values are optimistically biased. The leading predictors were Pull RFD (10.0%), CMJ power per body mass (7.2%) and eccentric CMJ peak force (6.6%). The risk tiers were HIGH 56 (44.4%), MEDIUM 39 (31.0%), and LOW 31 (24.6%). Conclusions: Multi-protocol force-plate data yield moderate internally cross-validated discrimination of injury risk. The near-perfect threshold metrics are apparent values, not generalisation performance. Rate of force development and eccentric force capacity outranked asymmetry indices. Prospective external validation is required before clinical use. Full article

The generation of Dead Hot Maps (DHMs) is an important part of PHENIX data analysis, as it helps reduce detector-related biases. During this step, detector units with abnormal behavior are identified, such as dead, hot, or extremely hot channels. Removing these channels from the analysis prevents detector effects from changing the measured collision distributions. This helps ensure that the final data sample represents the underlying collision physics more reliably. In this work, the standard DHM construction procedure is compared with anomaly detection methods that are commonly used in data analysis and data mining. These methods can offer simpler and faster alternatives to the conventional approach, with lower computational demand and less need for manual work. Five techniques were tested: Z-score analysis, Local Outlier Factor (LOF), One-Class Support Vector Machine (OCSVM), Isolation Forest (IF), and Kernel Density Estimation (KDE). The main goal was to examine which method gives a detector mask that is closest to the established statistical reference and is also physically reasonable. The results suggest that these anomaly detection approaches may be useful for detector monitoring and calibration tasks. Some of the tested methods are easy to apply, computationally efficient, and do not require special hardware. Because of this, they could be included in future automated detector quality control workflows. Their simple implementation also makes them suitable for use in different software environments. In this way, these methods may support faster and more consistent identification of anomalous detector towers. Full article

Poorly governed geoscience databases introduce subtle but systematic biases that propagate through mineral resource estimation workflows, distorting grade continuity and resource classification. While the importance of data quality in mining is widely acknowledged, quantitative demonstrations of how specific data management decisions translate into downstream technical and economic impacts remain limited. This study addresses that gap through three case studies derived from real-world industry examples. The first case study quantifies operator-dependent data extraction bias by comparing two independently generated datasets sourced from the same database on the same day. Despite nominal equivalence, the datasets differed materially in record counts and grade distributions, producing only negligible global volumetric differences (~0.1%) but up to ~5% local geometric variability and a 19.5% difference in estimated copper grade, resulting in a ~36% divergence in projected revenue. The second case study evaluates analytical uncertainty near detection limits using duplicate assay pairs, demonstrating that relative error increases markedly at low concentrations and that this behavior reflects inherent analytical limitations rather than laboratory non-compliance. The third case study examines the common practice of assigning detection limit placeholder values to unassayed intervals, showing that such substitutions can artificially generate ore in barren domains, whereas retaining null values and applying assignments during post-processing yields geologically and statistically coherent results. Collectively, these case studies demonstrate that hidden data biases can exert a stronger influence on resource outcomes than estimation methodology alone. The results highlight the need for standardized extraction workflows, explicit treatment of low-grade uncertainty, and validation practices that extend beyond global reconciliation metrics. Robust data governance and transparent, reproducible workflows are essential to reducing compounding uncertainty and improving confidence in mineral resource models. All datasets have been anonymized, scaled, or modified to prevent identification of any specific project, company, or operation. No confidential or proprietary datasets are disclosed. Full article

Precise estimation of aboveground biomass in urban forests is crucial for quantifying urban carbon stocks and supporting climate change mitigation efforts. However, the availability of allometric equations tailored for urban trees is limited. Existing equations often rely on data from harvested trees with restricted sample sizes and small diameters, thereby introducing substantial uncertainty into biomass assessments. This study utilized terrestrial laser scanning (TLS) in conjunction with a leaf-wood separation algorithm and a tree quantitative structure model (TreeQSM) as a non-destructive approach to develop new species-specific allometric models for four predominant evergreen broadleaved tree species in the urban forests of Shanghai, based on 10 sample plots and 303 trees. The results showed that TLS-derived multivariate models, which incorporated diameter at breast height (DBH), tree height, and crown diameter, consistently outperformed models that only included DBH. Compared with the TLS-derived biomass, the previously published models showed varying degrees of deviation. Notably, there was a substantial overestimation for Camphora officinarum, with a bias of +31.5%. In contrast, Elaeocarpus decipiens, Ligustrum lucidum, and Magnolia grandiflora demonstrated smaller underestimations, with biases of −9.4%, −0.9%, and −4.5%, respectively. These discrepancies were primarily attributed to the extrapolation beyond the calibration diameter at DBH ranges of the published equations. These findings highlight the critical need for urban-specific models. Because destructive harvesting was not feasible in the urban environment, the TLS-derived biomass estimates were not validated against destructively measured biomass. The equations developed in this study provide improved tools for estimating urban forest biomass and carbon accounting for the four studied species under the sampled conditions in Shanghai. Full article

Results of a clinical trial may misrepresent the target population. Inappropriate inclusion/exclusion criteria, selective recruitment, and differential participation may restrict the number of patients who consent to join the study. Frequently, they differ from the real-world population in terms of key outcome-related characteristics. Such biased selection may make the therapeutic agent seem more or less effective within a certain stratum. While block randomization and an intent-to-treat design can help to minimize confounding within the trial sample, differences may still exist with respect to the distribution of these variables in the population. To mitigate misleading conclusions, the current manuscript illustrates how to ‘population-adjust’ survival estimates in a clinical trial using the Beckett–Bayes expansion method. In contrast to other population adjustment methods, this novel approach involves substituting the probability of belonging to a specific sample stratum with the corresponding population data (i.e., population borrowing). As a regression-free technique, the method is not susceptible to model misspecification. A large clinical trial focusing on cardiovascular-related deaths in patients with hypertension is used as an example to illustrate the method. The proposed framework is promising and easy to implement when reliable population-level information is available. Full article

Data obtained from experiments and surveys in human social systems are inevitably influenced by systematic measurement errors, and network data are no exception. Despite the prevalence of error in social network data, current research often lacks rigorous estimation of its expected precision, which may lead to biased conclusions. Signed networks, which encode both positive and negative relationships, constitute an important component of network science, and conducting measurement error analysis on them can substantially enhance the accuracy of social network analysis. This paper proposes a set of error measurement tools based on the Expectation-Maximization (EM) algorithm, specifically designed to estimate errors in signed network data. We extend traditional experimental error estimation to the network domain, derive a general error estimation method for signed networks, and validate its scientific validity and practical utility through extensive simulation experiments on both synthetic and real-world networks. The experiments reveal that network density and the ratio of positive to negative edges significantly influence the posterior probability distribution of the adjacency matrix. Specifically, as density increases, edge estimation accuracy exhibits a U-shaped trend, and the proportion of negative edges shows a nonlinear relationship with accuracy. The proposed method is applicable to repeatedly measured signed networks and provides a reliable framework for reconstructing network structures as faithfully as possible. Full article

Playing position influences soccer match demands, but comparisons may be biased by unequal playing time and repeated observations from the same players. This retrospective observational study compared GPS-derived running and change-of-velocity demands across five positional groups while accounting for within-player clustering and match exposure. The primary analysis included 84 player-match observations from 21 male professional outfield players across six competitive matches. Linear mixed-effects models included playing position and match as fixed effects, log-transformed playing duration as a covariate, and player as a random intercept. After Holm correction, position was associated with total distance, high-intensity running distance at ≥19 km/h, distance at ≥25 km/h, sprint count, peak speed, accelerations at ≥3 m/s², and decelerations at ≤−3 m/s² (adjusted p ≤ 0.020). At equivalent exposure, central defenders recorded approximately 49% less high-intensity running than wingers and fewer sprints, accelerations, and decelerations than several other groups. Midfielders achieved a 3.06 km/h lower adjusted peak speed than side defenders/full-backs. Findings were robust to two sensitivity analyses. Position-specific monitoring should combine accumulated match dose with exposure-adjusted estimates and account for repeated observations when players contribute data from multiple matches. Full article

Multimodal fake news detectors may learn biased dependencies from imbalanced event distributions and incidental text–image associations, causing attention to capture dataset-specific patterns rather than reliable discriminative evidence. To address this problem, we propose an Attention-level Causal Intervention Framework (ACIM), which performs causal adjustment directly within the attention learning process. Unlike prior causal debiasing methods operating at the feature-representation level, ACIM intervenes in attention distributions where cross-modal bias emerges. By treating attention representations as mediators, ACIM applies front-door causal intervention to mitigate confounding effects and estimate attention-level intervention effects without requiring fully observed confounders. This principle is implemented through a Causal Attention Layer Module (CALM), integrated into BERT-based textual and Swin Transformer-based visual encoders to jointly model in-sample and cross-sample attention. A causal-aware fusion layer further reconstructs cross-modal attention to suppress misleading text–image co-occurrence patterns. Experiments on Twitter and PHEME achieve accuracies of 0.906 and 0.909, improving upon the strongest reported accuracy baselines by 0.9 and 0.6 percentage points, respectively, while maintaining competitive precision, recall, and F1 performance. Ablation and sensitivity analyses further support the contribution and stability of the proposed approach. Full article

Common prosperity places the distributional consequences of digitalization at the center of China’s current development agenda. This study develops a two-level analytical framework linking macro-level structures to household income positions. The empirical analysis combines a prefecture-level panel for 2009–2022 with the China Family Panel Studies to examine how city-level digital development relates to the urban–rural income gap. City-level estimates indicate that more advanced digital economies are associated with a wider urban–rural income gap during the sample period. The mechanism estimates are consistent with possible channels involving skill-biased technological progress and unequal digital access and absorptive capacity. Moderation tests indicate that interaction between an “enabling government” and an “efficient market” can mitigate this widening trend. Evidence from the Broadband China pilot further suggests that infrastructure expansion without complementary institutions may intensify polarization, highlighting the limits of policy intervention. At the micro level, digital participation exhibits an inclusive yet asymmetric pattern. Internet use is associated with lower income-based relative deprivation in both groups, with a slightly larger estimated reduction among urban households. Taken together, the evidence points to a distributional tension between digital expansion and equity and supports policies that pair connectivity with stronger capabilities and inclusive institutions. Full article

Background: Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder that manifests differently across sexes, potentially impacting symptomatology, diagnosis, and treatment. Nevertheless, these legitimate sex differences in ADHD are further shaped by multiple sources of bias that distort prevalence estimates and clinical recognition. Methods: This critical review synthesizes evidence on a variety of factors, including sampling bias, referral bias, DSM criteria, and diagnostic overshadowing, that may contribute to the under-recognition and misdiagnosis of ADHD in females. Results: Studies suggest that sampling and referral biases inflate male representation by overemphasizing disruptive behaviors and underrecognizing inattentive presentations in females. Item-level symptom endorsement differences indicate that several DSM-5 criteria map more closely onto male-typical behaviors, reducing sensitivity to female-specific manifestations. Diagnostic thresholds, age-of-onset, and pervasiveness criteria further disadvantage females, whose symptoms are often less overt or more context-dependent. Recognition bias and diagnostic overshadowing by mood and anxiety disorders contribute to missed or delayed diagnoses. Conclusions: Together, these findings highlight that observed sex differences in ADHD reflect not only true psychobiological variation but also systematic biases in assessment and diagnosis, underscoring the need for more sex-sensitive diagnostic frameworks and improved clinical awareness. The extent of these diagnostic disparities may also vary across cultural contexts and healthcare systems, including differences in symptom perception, referral pathways, and access to specialist assessment. Full article

We establish an asymptotic theory for the Jones inverse-weighted kernel density estimator when length-biased observations form a strictly stationary short-range dependent sequence. The statistical difficulty is intrinsically composite: reciprocal weighting is singular at the origin, the normalizing mean is estimated from the same dependent sample, kernel localization shrinks with the bandwidth, and the centered summands form a row-wise stationary triangular array whose envelope diverges at rate $h_n^{-1}$. Under a non-negative compactly supported Lipschitz kernel, an inverse-moment condition, geometric $\alpha$-mixing, local regularity of the target density, and uniform local bounds on lagged bivariate densities, we prove strong uniform consistency on compact subsets of $(0,\infty )$ and, separately, the uniform stochastic bound $O_{\mathbb{P}}\{h_n^2+{(\log n/\left(n{h}_n\right))}^{1/2}\}$. A covariance-localization argument shows that the scaled serial-covariance contribution is $O\left\{h_n\log (1/h_n)\right\}=o\left(1\right)$, so the first-order pointwise variance coincides with that of the corresponding independent length-biased estimator. Pointwise and finite-dimensional Gaussian limits are obtained by an explicit big-block/small-block argument with off-diagonal covariance control. The ratio normalization is treated directly: its variance contribution, its product with the localized fluctuation, and its cross-covariance with that fluctuation are all negligible at the $\sqrt{n{h}_n}$ scale. We further derive first-order AMSE and AMISE criteria, their oracle bandwidths, and feasible pointwise studentization under undersmoothing. The numerical study separates oracle from data-driven bandwidth selection, evaluates full-ratio HAC and moving-block corrections, examines a Frank-copula Markov robustness design, and benchmarks the Jones estimator against an alternative length-biased estimator. The simulations support the first-order theory while demonstrating that persistent short-range dependence can remain consequential for finite-sample uncertainty. Full article