Search Results (1,041)

The diagnosis of cervical spondylotic myelopathy (CSM) relies mainly on clinical symptoms and structural imaging, highlighting the need for objective functional biomarkers. This study investigated alterations in resting-state magnetoencephalography (MEG) functional connectivity in CSM and evaluated whether multiband weighted phase lag index (wPLI) features could distinguish CSM patients from healthy controls (HCs). Eyes-closed resting-state MEG data were acquired from 31 CSM patients and 32 HCs. Region-of-interest-level wPLI connectivity was calculated in the theta, alpha, beta, and gamma bands and used to train multiple machine learning classifiers. Model performance was assessed using nested group cross-validation, and SHapley Additive exPlanations (SHAP) were used to interpret the best-performing model. Patients with CSM exhibited frequency-specific connectivity alterations, particularly in the theta and gamma bands. Logistic regression achieved the best overall discriminative performance, and SHAP analysis indicated that classification was driven mainly by long-range theta-band connections and gamma-band connections involving the frontal pole. These findings suggest that CSM is associated with measurable reorganization of large-scale cortical networks and that resting-state MEG connectivity combined with explainable machine learning may provide a promising framework for exploring candidate neurophysiological biomarkers of CSM. Full article

Agile outcome models commonly map Jira variables directly to a retrospective label and then explain the prediction through fragmented feature attributions; they rarely separate domain concepts, team clustering, unresolved work, and concept-label coupling. This study evaluates a domain-informed, concept-bottleneck-style explainable AI architecture for retrospective diagnosis of Agile Epic outcomes. A frozen Jira export of 10,000 unique issue-level records was linked to a pre-specified analytical cohort of 180 Epics across 14 teams. Six experts rated efficiency, effectiveness, sustainability, and contextual risk, while outcomes were recorded as Successful, Challenged, or Unsuccessful. Because the outcome labels and concept ratings were informed by the same Jira evidence, the models estimate consistency with an expert labelling procedure, rather than independent project success. Under five-fold group-aware cross-validation, the fixed-configuration flat LightGBM achieved macro-F1 = 0.864 ± 0.053 and the fixed-configuration HMXAI/CBM-style model achieved 0.843 ± 0.084. These descriptive primary scores are not a joint nested-model-selection comparison. The proposed method, therefore does, not demonstrate a performance improvement; its contribution is an inspectable diagnostic structure. Performance fell materially on the resolved-only subset (LightGBM macro-F1 = 0.645), and model-specific nested, leave-one-team-out, calibration, uncertainty, correlation, and intervention analyses further bound the claims. Concept interventions were not uniformly monotone, so the concept layer is domain-interpretable in form but not yet user-validated as actionable. The study contributes a transparent audit of when concept-level diagnosis can complement flat classification and when circularity, censoring, and shortcut learning restrict interpretation. Full article

Retinal responses to ultrasound depend on carrier frequency, but whether a constrained tissue-to-neuron model can predict this dependence on a common published response scale remains unclear. We developed an acoustic-radiation-force (ARF) modelling workflow that couples an effective tissue-scale stress proxy to a stochastic Hodgkin–Huxley retinal ganglion cell population. One non-negative affine observation model, shared across frequencies, linked raw simulated spike counts to the published three-frequency ex vivo RGC response scale without frequency-specific rescaling. In a nested leave-one-frequency-out model-selection assessment, the workflow achieved an overall $Q^2$ of 0.915 and an RMSE of 0.0752 across 25 held-out observations. Mean-response transfer remained strong across frequencies, although probabilistic calibration was weakest at 1.9 MHz. An in-sample fixed-map sensitivity analysis using moment-matched log-normal, Gamma, and Weibull thresholds yielded overall RMSEs of 0.0434–0.0461 and preserved positive overall K-ablation NLPD differences of 0.219–0.238, with the effect concentrated at 43 MHz. These results show that shared-scale modelling captures the main cross-frequency response structure and that a high-threshold inhibitory functional component improves the description of high-intensity rolloff. Within the published three-frequency dataset, the workflow provides a reproducible benchmark for targeted comparisons of retinal ultrasound mechanisms. Full article

We explore the capability of a Large Language Model (LLM) to perform specific computations in mathematical physics: the task is to compute the coordinate Bethe Ansatz solution of selected integrable spin chain models. We select three integrable Hamiltonians for which the solutions were unpublished; two of the Hamiltonians are actually new. We observed that the LLM semi-autonomously solved the task in all cases, with a few mistakes along the way. These were corrected after the human researchers spotted them. The results of the LLM were checked against exact diagonalization (performed by separate programs), and the derivations were also checked by the authors. The Bethe Ansatz solutions are interesting in themselves. Our second model manifestly breaks left–right invariance, but it is PT-symmetric; therefore its solution could be interesting for applications in Generalized Hydrodynamics. And our third model is solved by a special form of the nested Bethe Ansatz, where the model is interacting, but the nesting level has a free fermionic structure lacking $U\left(1\right)$-invariance. This structure appears to be unique and it was found by the LLM. We used ChatGPT 5.2 Pro and 5.4 Pro by OpenAI. Full article

This paper proposes a unified invariant-set-based reliable control framework for Takagi–Sugeno (T–S) fuzzy systems subject to actuator saturation and faults. The considered model incorporates both matched actuator faults and mismatched external disturbances, which provides a more realistic control setting. To address these challenges, a unified control framework is developed to systematically account for input constraints and actuator fault effects. A sequence of nested invariant ellipsoidal sets, together with corresponding set-dependent control gains, are constructed to guarantee that state trajectories starting within the designed outer invariant sets progressively converge toward a minimized target set. Based on this structure, relaxed LMI-based conditions are derived to compute both the invariant sets and the associated control laws via convex optimization. Finally, numerical examples demonstrate the effectiveness of the proposed method. Full article

Robert Shiller proposed perpetual futures in 1993 to create derivative markets for assets that are illiquid or whose price cannot be observed directly. Cryptocurrency markets later built the instrument under a different funding rule. We give a single no-arbitrage result that nests both designs: the perpetual price is the present value of a benchmark flow discounted at the funding rate, so the funding rule fixes both the benchmark and the discount. A random time change represents the price as the expected spot at the first event of a clock whose intensity is the funding rate. This yields the main structural result, that stochastic volatility moves the basis only through the carry, so a volatility risk premium, and not volatility itself, can break the peg. We then read price discovery as nonlinear filtering in which the funding rule is a feedback observer whose gain is the funding intensity and the peg the fixed point of a stochastic approximation, and we give a segmented market equilibrium under which the pre-listing premium is structural rather than behavioral. In the June 2026 SpaceX market, the last pre-listing closes were $172.84 on Hyperliquid and $170.82 on Binance, compared with the listed equity’s $185 close on 18 June and the $135 bookbuilt offer. Simulation matches the pricing results to their closed forms. Generative Bayesian computation recovers the funding intensity sharply but not the softness of the anchor. Full article

Background/Objectives: Brazilian locally adapted sheep breeds represent valuable genetic resources. However, the commercial valorization of these breeds depends in part on genetic certification to support traceability and verify product origin. This study aimed to identify and evaluate a minimum set of highly informative SNPs for accurate and efficient breed assignment across five Brazilian locally adapted sheep breeds. Methods: A total of 677 samples were genotyped using the Embrapa Multispecies 65 K Illumina Infinium 1 chip, which contains 2926 markers for Ovis aries. The dataset was partitioned into training (n = 566) and independent testing (n = 111) sets. Markers were ranked according to genetic differentiation based on pairwise Wright’s fixation index (F_ST) using the Toolbox for Ranking and Evaluation of SNPs (TRES), generating three nested reduced panels of 288, 192, and 96 SNPs. Panel performance and preservation of population structure were evaluated using Random Forest classification, Principal Component Analysis (PCA), and ADMIXTURE. Results: The 96-SNP panel achieved classification accuracy comparable to that of the 2145-SNP post-QC baseline and the 192- and 288-SNP panels (Cochran’s Q test: Q = 6.00, df = 3, p = 0.112), with no statistically significant difference in classification performance despite the substantial reduction in marker number. PCA and ADMIXTURE analyses indicated that the 96-SNP panel preserved the major population structure observed with the full post-QC marker set. Conclusions: Although formal analytical validation of a dedicated low-density genotyping assay remains necessary before routine implementation, the in-silico marker selection and empirical validation performed here provide an evidence-based framework for translating high-density genomic information into accessible and cost-effective applications for breed assignment, traceability, and conservation of Brazilian locally adapted sheep genetic resources. Full article

Electric dynamometer test benches generate heterogeneous multi-channel logs whose structural completeness, physical consistency, and operating-condition support vary across files. We propose Coupling-Gated Data Quality Index (CG-DQI) sample weighting, a fold-local preparation protocol for 60 s next-window temperature-change prediction. Deployable weights use training-side structural and physical quality signals, while a target-magnitude coupling gate screens candidate association with the absolute target. The dataset contains 30,093 valid windows from 134 files and 30 date-defined campaigns. File-level leave-one-file-out validation reduced Ridge MAE from 0.7035 to 0.6932 °C and the mean MAE across three fixed Random Forest initializations from 0.4556 to 0.4489 °C per window. Strict nested leave-one-campaign-out analysis retained sparse CG-DQI in all 30 folds and yielded a campaign-level Ridge MAE difference of 0.0152 °C with a 95% bootstrap confidence interval (CI) of 0.0054–0.0309, while the 99th percentile absolute error (P99) increased by 0.0710 °C. Fixed-parameter XGBoost preserved a positive mean MAE direction, although its campaign interval crossed zero. CG-DQI, therefore, provides a traceable MAE-oriented preparation option within the present bench and campaign range; tail-sensitive and cross-device use requires an explicit constraint or recalibration. Full article

Under ISO 50001, energy performance is monitored through Energy Performance Indicators (EnPIs) and energy baselines. In practice, the energy-to-production ratio (kWh/t) is commonly adopted by default and validated solely by the coefficient of determination ($R^2$), which is insensitive to systematic bias, to the base load contained in the intercept, and to the residual structure that reveals an omitted explanatory variable. This work organizes well-established statistical and engineering checks into a sequential, four-outcome decision procedure anchored to the diagnosis of Significant Energy Uses (SEUs): retain the simple ratio, adopt a regression baseline with the same variable, switch to the SEU-associated variable, or reject the model as structurally misspecified. Relative to common practice, the procedure makes three methodological corrections explicit: in-sample NMBE is identically zero for OLS models with an intercept and is therefore defined out of sample; residual diagnostics are evaluated against exact, design-specific Durbin–Watson critical values with a Šidák-corrected family-wise error of 0.044–0.050 (versus ≈0.14 uncorrected); and the candidate-variable step uses a partial F-test on nested models, since the naive residual-versus-variable regression is attenuated by collinearity with production. The procedure is characterized on synthetic data with known truth ($N=1000$ replicates per cell): against an interannual drift of ≈2%/yr, its sensitivity reaches 1.00 at $n=72$ months while an $R^2$-only criterion has sensitivity 0.00, and with a base-load fraction of ≈0.28 the $R^2$-only rule retains the biased ratio in 100% of the replicates; specificity under a correct ratio is 0.95–0.96, and the adopted thresholds lie in a stable region of the ($R^2$, $f_0$) sensitivity sweep. The procedure is then demonstrated on six industrial cases; most notably, in a fuel oil power plant a pooled baseline with $R^2=0.998$ is rejected (Durbin–Watson $=0.79$ versus an exact critical value of 1.64; runs test $p<0.001$) because of drift in specific fuel consumption that $R^2$ cannot detect, and its out-of-sample validation over 37 rolling origins shows that an aggregated bias of $+0.31\%$ can mask an origin-to-origin drift from $-1.7\%$ to $+2.1\%$. The contribution is not a new indicator or a new statistic, but the integration of indicator selection and multistage statistical validation into a single auditable decision procedure whose operating characteristics are quantified. Full article

Emotion recognition in children’s drawings is difficult because affect is carried by sparse strokes, symbolic objects, and overall composition rather than by the stable appearance statistics of photographs. We built a reproducible four-class benchmark (Angry, Fear, Happy, Sad) on a single public corpus of 818 children’s drawings and compared three transfer-learning regimes under identical stratified five-fold splits with nested model selection: ResNet-50, ViT-B/16, and an end-to-end fine-tuned SigLIP image encoder (SigLIP-FT). SigLIP-FT reached the highest macro-F1 (0.773 ± 0.028), ahead of ViT-B/16 (0.700 ± 0.040) and ResNet-50 (0.598 ± 0.038), and was the best calibrated (ECE 0.119). Frozen-feature linear probes preserve this ordering (0.541, 0.648, 0.731), locating the advantage in the pretrained representations rather than in fine-tuning, while zero-shot SigLIP reaches only 0.510, so task-specific supervision remains necessary. Margin-based abstention raised retained-set macro-F1 to 0.810 at 78.0% coverage and 0.844 at 61.4% coverage—post hoc operating points computed on the pooled out-of-fold predictions; deployment thresholds must be fixed on independent data—and SigLIP-FT attains the lowest area under the risk–coverage curve (AURC 0.126 versus 0.199 and 0.278). Residual errors are highly structured: 87.0% lie within the negative-emotion triad, and Fear is the hardest category for all three architectures. All findings are established on this single corpus; their transfer to other collections is an open question. Coverage–performance behavior, rather than a single full-coverage score, is the appropriate reporting standard for ambiguous visual domains of this kind. Full article

The construction sector faces a critical need to minimize its carbon footprint, which is currently stimulating the development of geopolymer concrete using recycled coarse aggregates as an eco-friendly material compared with Portland cement. Accurate prediction of the compressive strength of this eco-efficient concrete is complex, however, as a result of the complex, non-linear interactions between many of the mix-design and curing parameters. Although modern scientific literature and engineering practices have increasingly adopted machine learning (ML) for concrete strength prediction, a significant scientific gap remains. Most existing studies rely on “black-box” models that lack sufficient interpretability and frequently overlook the severe risk of data leakage during validation, limiting their practical engineering application. To address this gap, this study proposes a robust, data-leakage-aware framework driven by a rigorous nested GroupKFold cross-validation strategy. By grouping concrete samples by their unique Mix_ID, this approach ensures genuine generalization to entirely unseen mixtures. Within this reliable validation scheme, the CatBoost algorithm is utilized for compressive-strength prediction, with the Dung Beetle Optimizer (DBO) serving as an effective tool for hyperparameter tuning. The evaluation results across multiple random seeds show that the DBO–CatBoost model significantly outperforms the default CatBoost, rigorously tuned baseline models (Support Vector Regression and Random Forest), and a comparative metaheuristic benchmark (PSO–CatBoost). It achieves the most stable distribution of errors and excellent predictive accuracy (Test ${\mathrm{R}}^2=0.9995\pm 0.0002$, RMSE = $0.3828\pm 0.0909$). In addition, the model predictions were demystified using the methods of SHapley Additive exPlanations (SHAP) and partial dependence plots (PDPs). The interpretability analysis revealed strong statistical associations, showing that Curing Time and Coarse Aggregate are the most prominent predictive features and the strongest pairwise interaction between each other; the NaOH molar concentration is the most important second-level influence on optimization of strength. Overall, the framework provides a robust data-driven screening tool that can assist in preliminary mix-design evaluation. By reducing the reliance on extensive empirical “trial and error” approaches, this predictive model supports more efficient material usage and facilitates preliminary optimization of low-carbon concrete formulations. Theoretically, this study advances the fundamental science of geopolymer materials by explicitly quantifying the complex, non-linear interactions between alkaline activators, curing conditions, and recycled aggregates. This provides a robust data-driven theoretical foundation for designing and optimizing next-generation eco-friendly concrete products and structures. Full article

Testing multinormality in two-level structural equation models (SEMs) presents a fundamental challenge because observations from the same level-2 unit are correlated, violating the independence assumption required by classical normality tests. In this paper, we develop a novel generalized Shapiro–Wilk ($GW$) test that explicitly accounts for this dependence. The proposed method rearranges the dependent observations into a random matrix and employs principal component analysis (PCA) to project this matrix onto a set of principal directions, achieving effective dimension reduction. On each projected direction, the scale-invariant Shapiro–Wilk statistic is applied to test for sphericity, leveraging the property that spherical distributions preserve the null distribution of such statistics. The Johnson $S_B$-transform is then used to approximate the null distribution of the combined test statistic. A Monte Carlo study demonstrates that the proposed $GW$ test controls type I error rates satisfactorily and exhibits strong power against a range of non-normal alternatives, including heavy-tailed and asymmetric distributions. The method is further illustrated using real alcohol use data from nested families, highlighting its practical utility. Comparative evaluation indicates that the $GW$ test performs favorably relative to a recently proposed approach. The procedure is applicable to balanced level-1 designs and provides researchers with a necessary diagnostic tool for assessing multinormality assumptions in two-level SEMs. Full article

Urban low-carbon transition is increasingly shaped by the interaction between digital infrastructure, intelligent applications, land-space allocation, and regional governance. Existing studies have mainly examined whether the digital economy or smart-city development can reduce emissions, but less attention has been paid to the coordination between digitalization and intelligentization, the carbon cost of digital infrastructure, and the spatial consequences of local gains. This research defines digital–intelligent integration as the coupling coordination between digitalization and intelligentization subsystems. Using panel data for 39 prefecture-level cities in the Middle Reaches of the Yellow River Urban Agglomeration from 2013 to 2022, it applies Global Moran’s I, a spatial Durbin model, partial-derivative effect decomposition, alternative spatial weight matrices, alternative dependent variable tests, and multidimensional heterogeneity analysis. The own-city coefficient of digital–intelligent integration in the carbon-efficiency model is positive (0.0282, p < 0.05), whereas the spatial-equilibrium direct effect is statistically insignificant. These quantities are not short- and long-run estimates: the former is a conditional model coefficient, while the latter incorporates spatial feedback. The indirect effect on neighboring carbon efficiency is negative and remains negative under contiguity, economic-distance, and geo-economic nested matrices. Under an otherwise identical fixed-effects specification, digital–intelligent integration lowers local construction-land carbon intensity but raises neighboring intensity. The structural estimates further show that local conversion is weaker in industrially and energy-intensive cities. Digital–intelligent integration should therefore be interpreted as a governance capacity rather than a net-carbon technology; its regional effect depends on industrial lock-in, infrastructure-energy demand, and cross-city responsibility sharing. Full article

Background: Spinal muscular atrophy (SMA) is mainly caused by biallelic SMN1 inactivation. While most patients carry homozygous deletions, 3–5% are compound heterozygotes, making molecular diagnosis challenging. Methods: A tiered diagnostic strategy was applied to 17 pediatric patients, combining copy number analyses (MLPA and targeted long-read sequencing, tLRS), sequence variant detection (RT-PCR cloning and sequencing, allele-specific long-range PCR with nested PCR, and tLRS), and structural variant analysis (ultra-long-read sequencing, Ultra-LRS). Results: Copy numbers were concordant between MLPA and tLRS. MLPA-suggested gene conversions were confirmed by tLRS, while discordant total copy numbers were resolved as large deletions by Ultra-LRS. RT-PCR cloning, and sequencing identified SMN1 variants in 11/12 cases and confirmed aberrant splicing in three cases, but failed for large deletions. AS-LR-PCR with nested PCR characterized the variants in 13/15 but failed in gene conversion cases. tLRS achieved definitive diagnosis in all cases, and Ultra-LRS precisely delineated breakpoint junctions of two large deletions. Conclusions: A hierarchical complementary strategy integrating copy number, sequence, and structural analyses is essential for the accurate diagnosis of compound heterozygous SMA. Full article

Flooding and farmyard manure (FYM) application trigger complex, non-linear redox reactions that govern micronutrient availability in calcareous soils, yet predictive modelling of these dynamics using machine learning (ML) remains largely unexplored, and the present study was designed to address this gap. To this end, seven supervised ML algorithms—Ridge Regression, Support Vector Regression (SVR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Gradient Boosting Machine (GBM), Artificial Neural Network (ANN), and Cubist—were compared under a unified nested cross-validation scheme to predict DTPA-extractable Fe, Mn, Cu, and Zn concentrations in a flooding incubation experiment comprising 10 contrasting calcareous soils (Entisol, Mollisol, Inceptisol, Vertisol) from the Atabey Plain (Isparta, Türkiye), two FYM doses, and five flooding durations (n = 100). Under Leave-One-Out Cross-Validation (LOO-CV), rule- and tree-based ensemble methods consistently outperformed linear and neural network models, with Cubist achieving the best performance for Fe (R² = 0.812) and Mn (R² = 0.915), XGBoost for Cu (R² = 0.929), and GBM for Zn (R² = 0.919). However, a stricter leave-one-soil-out (LOSO) validation with grouped inner cross-validation revealed that this accuracy is element-specific in its transferability: Mn predictions remained robust on previously unseen soils (R²cv = 0.739) and Fe moderate (R²cv = 0.412), whereas Cu and Zn did not generalise beyond the soils used for training, indicating that their high within-soil accuracy reflects soil-specific rather than transferable structure. SHAP analysis revealed that flooding duration was the dominant predictor of Fe and Mn availability, amorphous Fe oxide content was the primary driver for Cu, and plant-available phosphorus (Olsen-P) was the principal feature for Zn. These findings demonstrate that combining ensemble ML with SHAP interpretability enables element-specific, cross-soil-validated and mechanistically interpretable prediction of micronutrient dynamics under varying redox and organic amendment conditions, while highlighting cross-soil transferability as a critical consideration for deploying such models in calcareous agroecosystems. Full article