Search Results (390)

High-volume, long-duration flowback in ultra-deep fractured wells couples pressure, rate, water production, fracture conductivity, and stress-sensitive reservoir properties, making it difficult for pressure transient analysis (PTA), flowing material balance (FMB), and rate transient analysis (RTA) to maintain parameter continuity across flowback stages. This study proposes a physics-constrained artificial-intelligence (AI)-assisted workflow centered on FMB. PTA provides permeability, fracture half-length, and fracture-conductivity priors; RTA, Blasingame, and Agarwal–Gardnerdiagnostics provide production-dynamic constraints; and machine-learning models perform anomaly screening, stage recognition, time-series correction, type-curve discrimination, and multi-method fusion. The workflow was applied to Well Baitan 1, an ultra-deep fractured gas well with eight-stage fracturing and multi-regime flowback data. Isolation forest preprocessing removed 32 abnormal records, random forest (RF) drainage-type classification reached 93% accuracy, and long short-term memory (LSTM) correction improved production-forecast fitting from 87% to 95%. Neural-network fusion yielded matrix permeability of 0.49 mD and dynamic reserves of 1.93 × 10⁴ m³, reducing static geological-volume validation error to 2.8%. The results show that AI improves productivity evaluation when constrained by diagnostic flow models and geological validation, providing a traceable basis for optimizing flowback intensity, monitoring frequency, and stabilized deliverability estimation. Full article

Transient contact responses in confined guide channels contain sharp events and parameter-dependent phase shifts, which make fixed-output full-history surrogates difficult to train. This study develops a time-conditioned residual surrogate (TC-ResNet) to propagate uncertainty in the bounded center-of-mass eccentricity components into the transient acceleration responses of a generic pressure-driven, high-overload axisymmetric body. The nonlinear reference model includes prescribed base pressure, wall contact and impact, velocity-dependent friction, gravity, pitch and yaw, and eccentricity. TC-ResNet predicts one response value for each parameter–time query by combining normalized physical parameters with Fourier-embedded time. The axial acceleration is modeled directly, whereas low-frequency radial trends and sliding root-mean-square (RMS) curves represent dominant lateral motion and local vibration intensity. On a common 108-case test set, TC-ResNet achieved the highest coefficient of determination (R²) for a_x^low (0.850), a_x^rms (0.839), and a_z^low (0.801), as well as the highest macro-mean R² (0.848). The fixed-output multilayer perceptron (MLP) remained best for a_y (0.995) and a_z^rms (0.775), demonstrating that the proposed model is not uniformly superior. Interval analysis shows that the prescribed pressure load limits axial sensitivity, whereas radial offsets alter eccentric pressure moments and wall contact; the axial offset primarily changes contact and friction moment arms, and inclination mainly affects the later radial response through gravity decomposition and accumulated contact differences. Accuracy approaches a plateau between 378 and 504 training cases. On the same central processing unit (CPU), TC-ResNet requires 0.081 s per curve (approximately 170× faster than the reference solver), and graphics processing unit (GPU) inference requires 0.019 s per curve. The reported envelopes support qualitative sensitivity analysis, but experimental calibration and validation remain necessary. Full article

Background and Objectives: The C-reactive protein-to-albumin ratio (CAR) is associated with mortality and poor outcome after acute ischemic stroke, but the association is not the same as added clinical usefulness. We evaluated whether admission CAR and follow-up CAR provide prognostic information beyond a prespecified parsimonious clinical reference model comprising age, neurological severity, and admission glucose in critically ill patients with acute ischemic stroke. Materials and Methods: In this single-center retrospective cohort of 146 adults with acute ischemic stroke managed in intensive care, CAR was calculated from C-reactive protein and serum albumin at emergency department admission and at the first intensive care laboratory assessment. The primary outcome was 90-day all-cause mortality. A clinical reference model (age, admission National Institutes of Health Stroke Scale [NIHSS] score, admission glucose) was compared with the same model augmented by log-transformed CAR using the area under the receiver operating characteristic curve (AUC), the DeLong test, likelihood-ratio (LR) testing, and bootstrap optimism-corrected performance. Results: Ninety-day mortality occurred in 32 of 146 patients (21.9%) (full cohort; primary complete-case analysis: 141 patients with 31 events). In the primary complete-case analysis (n = 141; 31 events), admission CAR was not independently associated with mortality (odds ratio per 1-SD log CAR 1.46, 95% confidence interval 0.91–2.33; p = 0.115). Adding admission CAR changed the AUC from 0.768 (0.676–0.860) to 0.783 (0.693–0.874) (ΔAUC +0.016, 95% confidence interval −0.024 to +0.055; DeLong p = 0.444; LR p = 0.113), with optimism-corrected point estimates of 0.750 and 0.754. Follow-up CAR did not add value (LR p = 0.159), and change in CAR did not improve discrimination (ΔAUC +0.001; LR p = 0.945); because intensive care sampling times were not standardized, these analyses assess incremental prognostic information rather than CAR kinetics. Admission CAR was not associated with poor 90-day functional outcome (odds ratio 1.09, 95% confidence interval 0.83–1.44; p = 0.530). Conclusions: Admission CAR did not demonstrate measurable incremental prognostic value beyond the prespecified clinical reference model, and follow-up CAR and change-based analyses did not improve prediction, although non-standardized sampling times mean that serial CAR kinetics were not fully evaluated. The confidence intervals remain compatible with both no effect and a positive effect of uncertain clinical relevance, but the observed improvement was insufficient to support CAR as a stand-alone or routinely additive prognostic marker in this setting. Full article

Background: Pediatric head trauma is a common reason for emergency department presentation. Although most children have minor injuries, a small proportion harbor clinically important traumatic brain injuries requiring urgent intervention. Artificial intelligence (AI) may offer structured support in computed tomography (CT) decision making, but evidence regarding the performance of general-purpose large language models in pediatric head trauma remains limited. Objective: To evaluate the association between AI-based cranial CT recommendations and clinically meaningful outcomes in pediatric patients with blunt head trauma and to assess the diagnostic performance and clinical utility of the model. Methods: This retrospective single-center observational study included pediatric patients younger than 18 years with blunt head trauma who underwent cranial CT imaging and had complete outcome data. A general-purpose large language model generated binary CT recommendations (“CT recommended” or “CT not recommended”) using structured clinical information available at the time of emergency department presentation. The primary outcome was a composite adverse clinical outcome defined as the occurrence of at least one of the following: emergency surgical intervention, intensive care unit admission, intubation, neurological sequelae or mortality. Diagnostic performance metrics, calibration analysis and decision curve analysis were performed. Results: A total of 819 pediatric patients were included, and the AI model recommended CT in 530 patients (64.7%). The primary outcome occurred in 143 patients (17.5%) and was significantly more frequent in the CT-recommended group than in the CT-not recommended group (24.5% vs. 4.5%; OR 6.90, 95% CI 3.82–12.45; p < 0.001). Abnormal CT findings, emergency surgery, intubation and neurological sequelae were also significantly more common in patients for whom CT was recommended by the AI system. For the primary outcome, the AI recommendation demonstrated a sensitivity of 90.9%, specificity of 40.8%, positive predictive value of 24.5% and negative predictive value of 95.5%. Calibration analysis showed acceptable agreement between predicted probabilities and observed event rates. Decision curve analysis demonstrated greater net benefit than both the “treat-all” and “treat-none” strategies across a range of threshold probabilities. Conclusions: In this clinically selected cohort of pediatric patients with blunt head trauma who underwent cranial CT imaging, AI-based CT recommendations were strongly associated with adverse clinical outcomes and demonstrated high sensitivity and negative predictive value for identifying children at risk of clinically important events. These findings suggest that, within a clinically selected cohort of children who underwent cranial CT imaging, AI-generated CT recommendations were associated with clinically meaningful outcomes. However, these results should not be interpreted as validation of CT decision making in the broader pediatric head trauma population and require prospective validation in unselected cohorts. Full article

Background/Objectives: Delirium is prevalent among intensive care unit (ICU) patients and is associated with prolonged ventilation, longer ICU stays, increased mortality, and post-discharge cognitive impairment. Because most current prediction tools were developed using European cohorts, their transportability to other ICU settings remains unclear. The device support burden has received limited attention, and preprocessing data leakage remains a secondary methodological concern. Methods: This prospective study enrolled consecutive adults from the Internal Medicine ICU of Ankara Etlik City Hospital. Fifty predictors from electronic health records were documented at ICU admission or within the first 24 h. Device support variables were assessed as a distinct domain using ablation analysis. We developed an elastic net regularized logistic regression model and validated it using 5 × 4-fold nested cross-validation. Preprocessing was restricted to the training folds, and gradient boosting was used as a comparison method. We also evaluated a fixed exploratory 11-predictor reduced model. Results: Delirium was detected in 50.8% of patients. The full regularized model achieved an area under the receiver operating characteristic curve (AUROC) of 0.696 (95% CI: 0.613–0.767). The fixed exploratory reduced model had a higher apparent AUROC of 0.756 (95% CI: 0.684–0.819) than the other models. This apparent advantage was absent when feature selection was repeated within each outer training fold (nested reduced model AUROC 0.700). The full-model calibration slope was 0.741 (95% CI: 0.345–1.265). This confidence interval was wide and included the value of 1.0. The fixed reduced-model slope was 0.235 (95% CI: 0.068–0.886), indicating that the predicted probabilities were too extreme. Ablation analysis showed that device support variables added predictive value to clinical and laboratory variables, whereas environmental variables alone did not significantly discriminate between the two groups. Conclusions: In this single-center Turkish ICU cohort, regularized logistic regression showed moderate performance in predicting delirium. Nested feature selection did not reproduce the fixed reduced model’s apparent discrimination advantage. Device support variables are best interpreted as markers of care complexity rather than direct causal factors. External validation and recalibration are required before clinical use. Because delirium onset and device support start times were unavailable, the model estimated delirium risk across the ICU stay from predictors documented at ICU admission or within the first 24 h and was not time-anchored. Full article

Commercial greenhouses are becoming significant, controllable, and weather-dependent electricity loads in regions pursuing controlled environment agriculture (CEA). Their electrical demands—such as supplemental lighting, heating, ventilation/cooling, irrigation pumps, and CO₂ enrichment—exhibit substantial intra-day flexibility. This flexibility stems from the fact that plant productivity depends on time-integrated agronomic variables (e.g., Daily Light Integral, accumulated thermal time, and mean vapor pressure deficit) rather than instantaneous environmental setpoints. However, existing studies have predominantly focused on greenhouse thermal modeling and energy conservation, while decision-making models that integrate crop physiological characteristics into day-ahead (DA) market bidding remain limited. To bridge this gap, this paper develops a scenario-based stochastic DA bidding framework for a load aggregator, representing multiple smart greenhouses in a wholesale electricity market. In the DA stage, the aggregator submits hourly demand–price bidding curves based on price-conditional schedules of heating and lighting demand while satisfying coupled thermal–photon balance constraints. Fifty representative scenarios generated from 2023 to 2024 historical price and weather data through cGAN-based scenario generation and K-means scenario reduction are used for stochastic bid construction and feasibility analysis. A separate 30-day historical dataset from January 2025 is used for benchmark comparison. The aggregator portfolio consists of 100 greenhouses divided into five LAI-based crop groups, with 20 greenhouses in each group. Relative to the 15–25 °C trapezoidal temperature baseline, which is adopted as the primary practical benchmark, the proposed strategy reduces the mean daily electricity procurement cost by 15.80%. A reduction of 18.89% is also observed relative to the rigid 20 °C thermostat case, which is retained as a capacity-intensive reference. These results represent simulation-based operating-cost comparisons under a common equipment configuration and do not include equipment capital costs. Full article

The aim of structural reliability assessment (SRA) is to guarantee the safety, durability, and performance of structures; however, traditional methods like stochastic finite element analysis (SFEA) can be computationally prohibitive to use in practical situations. This paper introduces a novel framework for SRA utilizing deep neural networks (DNNs) implemented in an open-source program called TRA-DNN, replacing the resource-intensive finite element (FE) analysis with a DNN model. The DNN is trained using 6874 FE column models, including factors like geometric imperfections, resulting in a training database with 419,314 data records. It accurately predicts axial load-deformation curves for corroded steel CHS columns, enabling the determination of the ultimate capacities for columns with varying properties. The model’s accuracy is confirmed through rigorous quantitative and qualitative validation, including various failure modes. TRA-DNN employs the DNN model to perform SRA via Crude Monte Carlo Simulation (MCS), yielding results that are in high agreement with conventional SFEA, yet with significantly reduced computational time (1,388,250 times faster). In addition, TRA-DNN can be used to estimate the service life of CHS columns considering both corrosion propagation and load increase with time. Future research can utilize TRA-DNN to optimize column design and maintenance to minimize both risk and cost. Full article

Background/Objectives: Lactate is a key biomarker of tissue hypoperfusion in severe trauma. While admission lactate is widely used, the prognostic value of delayed lactate clearance remains a subject of ongoing investigation, notably due to the inherent risk of survivorship bias. This study assessed the association between lactate kinetics and in-hospital mortality in a cohort of severely injured patients admitted to the intensive care unit. Methods: We conducted a single-center retrospective cohort study including adult patients (≥18 years) admitted to the ICU for severe trauma between January 2018 and December 2022. Arterial lactate was measured at H0, H4 (±1 h), and H12 (±1 h). Lactate clearance (LC) was calculated using the formula: LC (%/h) = [(Lactate_t1 − Lactate_t2)/Lactate_t1] × (1/Δt) × 100, where positive values indicate decreasing lactate (clearance) and negative values indicate worsening lactatemia. LC was calculated over H0–H4, H0–H12, and H4–H12. Among the 38 patients who died before 12 h and were excluded from the LC H0–H12 analysis, non-survivors were significantly older, had higher severity scores, and presented more frequently with hemorrhagic shock. Predictive performance was evaluated using ROC curves and multivariate logistic regression. The composite prognostic score was assessed using bootstrap internal validation (1000 resamples) to provide an optimism-adjusted AUC estimate. Results: Among 318 patients (median age 36 years; 86.5% male), hyperlactatemia (>2.2 mmol/L) was present in 70.1% at admission and persisted in 39.9% at 12 h. Non-survivors exhibited higher lactate levels and lower LC at all time points. LC H0–H12 demonstrated the best predictive performance for in-hospital mortality (AUC = 0.75; 95% CI [0.69–0.81]; p < 0.001). For early mortality (≤48 h), LC H0–H12 achieved an AUC of 0.80 (95% CI [0.71–0.89]; p < 0.001). A composite prognostic score incorporating age >60 years, GCS ≤ 7, prothrombin time ≤ 55%, pH ≤ 7.29, and LC H0–H12 > −2.93 %/h demonstrated good discrimination (AUC = 0.84; optimism-adjusted AUC = 0.82). Conclusions: Lactate levels and 12 h lactate clearance are valuable prognostic markers in severe trauma. Given the inherent survivorship bias affecting the LC H0–H12 analysis, its prognostic performance should be interpreted with appropriate caution and within a multimodal clinical assessment. The proposed composite score is promising but requires prospective external validation before clinical implementation. Full article

High-speed railways are expanding into seismic zones, where earthquakes severely threaten the safety of vehicles on bridges. Existing seismic studies of high-speed railway bridges mainly address the structural fragility of the overall bridge, piers, and bearings. However, field observations show vehicles often derail before major bridge damage during earthquakes. Fragility analysis based on vehicle derailment as the failure criterion is urgently needed. A core task in fragility analysis of vehicle derailment is to select an optimal ground motion intensity measure (IM). In this study, the dynamic analysis of the vehicle–bridge interaction system under earthquakes was conducted using the incremental dynamic analysis (IDA) method. The characteristics of the IDA curves of the derailment coefficient and the wheel unloading were investigated. Ground motion IMs are classified into time-history-related and spectrum-related categories. The derailment coefficient and wheel unloading are adopted as engineering demand parameters (EDPs). Probabilistic seismic demand models (PSDMs) are established, and 22 candidate IMs are evaluated based on efficiency, practicality, and proficiency criteria. The results revealed that the proficiency of spectrum-related IMs is better than that of time-history-related IMs. The Sa(T₁) performs the best among the 22 candidate IMs for the derailment coefficient, and ASI performs the best for the wheel unloading. Full article

Background/Objectives: Deciding between active surveillance and treatment for prostate cancer patients often requires accurate risk assessment of prostate tumors detected with multi-parametric MRI. Conventionally, radiologists visually inspect MRI and use scoring procedures such as PI-RADS to help assess the scans. More recently, artificial intelligence (AI) applied to MRI has allowed for supplementation and is complementary to clinical assessment. However, AI is computationally expensive and severely saps scarce energy and water resources and requires special processing components, requiring alternate approaches that require less computation and fewer resources. The novel, simpler spectral/statistics approach that mimics color vision was previously successfully applied in a number of retrospective pilot studies of bi-parametric MRI of prostate cancer. The novel approach needs far fewer resources, is less computationally intensive, and is simpler than artificial intelligence to evaluate prostate tumors. However, these earlier spectral/statistics pilot studies required intervention by an analyst and too much time for implementation in future large patient studies that are needed to validate the novel approach. This retrospective pilot study further developed, applied, and tested new automation tools to expedite simpler spectral statistical techniques that need fewer resources to evaluate prostate tumors on multi-parametric MRI. Methods: Automated spatial registration, automated prostate organ segmentation, automated blob generation and selection for spectral signatures derived from the apparent diffusion coefficient, high-B-value DWI, and T2 MRI were performed on 76 consecutive patients in the PI-CAI cohort in this retrospective pilot study. The signal-to-clutter ratio (SCR) was computed using target signatures and the processed statistical metrics of the registered prostate bi-parametric MRI. The processed SCR, spectral/spatial features of blobs and clinical metrics predict clinically significant prostate cancer using multivariate logistic regression. The proposed method was assessed using the area under the curve (AUC) from the receiver operating characteristic curve. Results: AUC values of >0.90 were achieved by combining the SCR with blob and clinical metrics. Increasing the number of non-congruent, independent variables resulted in higher AUC scores. Restricting analysis to blob volumes > 0.1 cm³ achieved higher AUC values. The additional total savings in time by applying the new automation tools reduced the processing time by 80 to 170 min for 10 patients. Implementing the new automation tools resulted in an overall processing time of 40 to 80 min per 10 patients. Conclusions: Automating the spectral/statistics approach resulted in AUCs not inferior to those obtained from AI. The automation achieved sufficiently high AUCs and also reduced processing times, warranting future assessments in large patient cohorts. Full article

Background/Objectives: Alzheimer’s disease (AD) is a progressive neurodegenerative disorder necessitating accurate and timely diagnosis for effective clinical intervention. While deep learning methods have shown promise in AD classification, many rely on computationally intensive architectures and high-dimensional feature representations. This study introduces a lightweight hybrid framework combining deep feature extraction, dimensionality reduction, and adaptive classification for MRI-based Alzheimer’s disease stage classification. Methods: Utilizing MRI images from a publicly available Alzheimer’s disease dataset encompassing four clinical stages (Non-Demented, Very Mild Demented, Mild Demented, and Moderate Demented), deep features were extracted using a pre-trained SqueezeNet model as a fixed feature extractor, generating 1000-dimensional feature vectors. Due to the computational complexity and for the improvement of the model efficiency, the dimensionality reduction technique, Principal Component Analysis (PCA) was then applied. This resulted in an optimum representation of 100 principal components, retaining about 96% of the variance. Then, the performances of various machine learning classifiers such as k-Nearest Neighbors (kNN), Support Vector Machine (SVM), Decision Tree (DT), Neural Network (NN), Naïve Bayes (NB), Logistic Regression (LR) and Enhanced Fuzzy Min–Max Neural Network (EFMM) were tested. The accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (AUC), and confusion matrices were used to evaluate the performance. Stratified 5-fold cross validation was used to ensure the strength of our results. Results: The findings show that PCA has a significant improvement in classification accuracy for most of the models. In particular, the EFMM classifier outperformed the other classifiers, with an accuracy of 97.19% on the independent test set. After PCA, the AUC values for classes such as Mild Demented, Moderate Demented, Non-Demented and Very Mild Demented were obtained as 97.12%, 99.97%, 93.79% and 95.26% respectively. We further validated our proposed framework using stratified 5-fold cross validation which further corroborated the robustness of our proposed framework. The EFMM achieved a mean accuracy of 98.38% ± 0.36 and a mean macro-F1 score of 98.48% ± 0.43. Friedman statistical testing demonstrated that there were significant differences between the performance of the classifiers evaluated (p < 0.001), which further validated the performance of the EFMM. Conclusions: To sum up, the proposed SqueezeNet–PCA–EFMM is an effective and efficient method for Alzheimer’s disease stage classification under MRI images. The combination of SqueezeNet, PCA, and EFMM—led not only to high classification performance, but also to good cross validation results. Furthermore, this property of incremental learning is the intrinsic one of the EFMM and renders this framework interesting for its incorporation in the next-generation intelligent clinical decision supports in particular, as medical care evolves. Full article

Background/Objectives: Sepsis and septic shock remain leading causes of morbidity and mortality in intensive care units (ICUs) worldwide. While activation of the hypothalamic–pituitary–adrenal (HPA) axis is a vital adaptive response, the dissociation between cortisol and adrenal androgens like dehydroepiandrosterone (DHEA) is hypothesized to correlate with poor outcomes. This study aimed to evaluate the prognostic value of serum cortisol, DHEA, and the Cortisol/DHEA ratio in patients with sepsis admitted to the ICU. Methods: Clinical and laboratory data from 106 adult sepsis patients in a tertiary care hospital were retrospectively analyzed. Disease severity was assessed using Sequential Organ Failure Assessment (SOFA), Acute Physiology and Chronic Health Evaluation II (APACHE II), and Glasgow Coma Scale (GCS) scores. Baseline cortisol and DHEA levels were measured at the time of diagnosis. Logistic regression and Cox proportional hazards models were employed to identify predictors of in-hospital mortality and time to death, respectively. Results: A total of 106 adult sepsis patients with a mean age of 73.63 ± 13.70 years (55 female, 51.9%) were included in the study population. In the final logistic regression model, higher SOFA score was associated with greater odds of in-hospital mortality Odds ratio (OR) = 1.39 per 1-point increase, 95% Confidence interval (CI): 1.15–1.69, p < 0.001). Higher platelet count (PLT) was associated with lower odds of mortality (OR = 0.958 per 10 × 10⁹/L increase, 95% CI: 0.919–0.998, p = 0.042), whereas higher urea was associated with greater odds of mortality (OR = 1.112 per 10 mg/dL increase, 95% CI: 1.020–1.212, p = 0.016). The model area under the curve (AUC) was 0.7956. In univariable Cox analysis, higher baseline DHEA was associated with a greater hazard of death (Hazard ratio (HR) = 1.03, 95% CI: 1.014–1.046, p = 0.0002). In the final multivariable Cox proportional hazards model, baseline cortisol remained independently associated with the hazard of death (HR = 1.014, 95% CI: 1.007–1.022, p < 0.001), whereas DHEA was not retained. Notably, the Cortisol/DHEA ratio did not demonstrate independent prognostic value in either modeling approach. Conclusions: Higher baseline DHEA levels were associated with an increased hazard of death in univariable Cox proportional hazards analysis but did not retain independent prognostic significance in the final multivariable Cox proportional hazards model. While SOFA score remains the central prognostic tool, integrating continuous markers such as platelets and urea enhances risk stratification. The Cortisol/DHEA ratio offers no incremental prognostic utility over established clinical parameters. Full article

This study investigated the effects of controlled burns on biomass dynamics and fire behavior in areas of stricto sensu grassland in southern Brazil. The experiment consisted of conducting controlled burns in three experimental plots divided into five subplots, with data collection on fire behavior (rate of spread, fireline intensity, residence time, and heat released) followed by monitoring of vegetation recovery over 19 months. To assess the influence of fire on biomass increase, the Gompertz model was fitted, correlating the Absolute Growth Rate (AGR), the time to peak productivity and the time at which the growth curve reaches 95% of the asymptote with the fire behavior variables. The results demonstrated that low-intensity controlled burns were effective in reducing fuel load. Correlation analysis indicated that fire behavior variables had little influence on the rate of vegetation recovery, suggesting that the ecosystem tolerates low-severity disturbances well. The Gompertz model showed a satisfactory fit (R² = 0.86 for total biomass), proving to be a useful tool for management. It is concluded that the use of low-intensity prescribed burns is a viable and safe practice for protected areas. Full article

Karst regions face acute conflicts among production, living, and ecology (PLE) functions under the constraints of rugged terrain and rocky desertification. Existing studies have separately examined driving factors, interaction directions and intensities, and nonlinear thresholds, while how these dimensions co-evolve across space and time remains unclear. To address this gap, we used seven concentric buffers radiating from the built-up area as a spatial proxy for human activity intensity and constructed a framework within each buffer using Geodetector to identify core driving factors, Pearson correlation to quantify the direction and intensity of three pairwise interactions, and Pareto frontier analysis to capture nonlinear thresholds and tipping points. In the Guizhou-Guangxi karst region (2010–2019), the core drivers of production and ecology functions shifted from construction land through elevation to precipitation, while population consistently drove the living function. Along this gradient, the coupling relationships fluctuated in the core but improved in the periphery. Although trade-off intensities eased beyond 10 km, three Pareto frontier curves shifted from inverted-U patterns to monotonic trade-offs, indicating that this improvement was only quantitative while the interaction structure degraded. Tipping points disappeared in these curves, eroding the carrying-capacity buffers where win-win synergies remained attainable. Averaged correlation coefficients obscured this contrast between apparent improvement and structural degradation, and linear correlation analysis alone would not have detected it. These findings highlight the need for zone-specific management in karst regions, where monitoring based on Pareto-derived structural indicators can provide early warning of coupling degradation that averaged correlations mask. Full article

This study investigates the relationship between economic growth, trade openness, energy intensity, and carbon dioxide (CO₂) emissions in South Africa within the Environmental Kuznets Curve (EKC) framework over the period 1970–2022. Using quarterly time series data and the Autoregressive Distributed Lag (ARDL) modelling approach, the study examines both the short-term and long-term dynamics between economic activity and environmental degradation. Descriptive statistics, correlation analysis, unit root tests, ARDL bounds testing, error-correction modelling, Granger causality analysis, and diagnostic tests were employed to ensure robust empirical results. The Augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests indicate that all variables are integrated of order one, I(1), thereby satisfying the conditions for ARDL estimation. The ARDL bounds test confirms the existence of a long-term cointegrating relationship among carbon emissions, economic growth, trade openness, and energy intensity. The long-term results reveal a statistically significant negative coefficient for economic growth and a positive coefficient for the squared income term, indicating a U-shaped relationship between income and carbon emissions. Consequently, the conventional Environmental Kuznets Curve hypothesis is not supported for South Africa. The findings suggest that economic growth initially reduces environmental degradation; however, beyond a certain income threshold, further economic expansion increases carbon emissions. Trade openness and energy intensity exert positive and statistically significant effects on carbon emissions in the long run, implying that increased integration into global markets and continued dependence on energy-intensive production contribute to environmental degradation. The Error-Correction Model (ECM) reveals a negative and highly significant adjustment coefficient (−0.928), indicating that approximately 92.8% of short-term disequilibrium is corrected within one period. Granger causality results further show a unidirectional causal relationship running from trade openness to carbon emissions, while no significant causal relationship is found between economic growth and carbon emissions. The study concludes that economic growth alone is insufficient to achieve environmental sustainability in South Africa. Policy efforts should therefore focus on promoting renewable energy adoption, improving energy efficiency, strengthening environmental regulations, encouraging cleaner production technologies, and integrating environmental considerations into trade and industrial policies. These measures are essential for achieving sustainable economic development while meeting national climate-change-mitigation objectives. Full article