Search Results (294)

Background/Objectives: Wet age-related macular degeneration (AMD) is a progressive retinal disease characterized by macular neovascularization (MNV). Currently, the standard treatment for wet AMD is intravitreal anti-VEGF administration, which aims to control disease activity by suppressing neovascularization. In clinical practice, the decision to continue or discontinue treatment is largely based on the presence of fluid on optical coherence tomography (OCT) and changes in visual acuity. However, discrepancies between anatomic and functional responses can occur during these assessments. Methods: This article presents an artificial intelligence (AI)-based classification model developed to objectively assess the response to anti-VEGF treatment in patients with AMD at 3 months. This retrospective study included 120 patients (144 eyes) who received intravitreal bevacizumab treatment. After bevacizumab loading treatment, the presence of subretinal/intraretinal fluid (SRF/IRF) on OCT images and changes in visual acuity (logMAR) were evaluated. Patients were divided into three groups: Class 0, active disease (persistent SRF/IRF); Class 1, good response (no SRF/IRF and ≥0.1 logMAR improvement); and Class 2, limited response (no SRF/IRF but with <0.1 logMAR improvement). Pre-treatment and 3-month post-treatment OCT image pairs were used for training and testing the artificial intelligence model. Based on this grouping, classification was performed with a Siamese neural network (ResNet-18-based) model. Results: The model achieved 95.4% accuracy. The macro precision, macro recall, and macro F1 scores for the classes were 0.948, 0.949, and 0.948, respectively. Layer Class Activation Map (LayerCAM) heat maps and Shapley Additive Explanations (SHAP) overlays confirmed that the model focused on pathology-related regions. Conclusions: In conclusion, the model classifies post-loading response by predicting both anatomic disease activity and visual prognosis from OCT images. Full article

Extremely high temperatures can severely impact urban livability and public health safety. However, risk assessments for high temperatures in cold-region cities remain inadequate. This study focuses on Dalian, a coastal city in northeastern China. Utilizing multi-source data, we established a population density prediction model based on the random forest algorithm and a heat vulnerability index (HVI) framework following the “Exposure-Sensitivity-Adaptability” paradigm constructed using an indicator system method, thereby building a high-temperature risk assessment system suited for more refined research. The results indicate the following: (1) Strong positive correlations exist between nighttime light brightness (NL), Road Density (RD), the proportion of flat area (SLP), the land surface temperature (LST), and the population distribution density, with correlation coefficients reaching 0.963, 0.963, 0.956, and 0.954, respectively. (2) Significant disparities exist in the spatial distribution of different criterion layers within the study area. Areas characterized by high exposure, high sensitivity, and low adaptability account for 13.04%, 8.05%, and 21.44% of the total area, respectively, with exposure being the primary contributing factor to high-temperature risk. (3) Areas classified as high-risk or extremely high-risk for high temperatures constitute 31.57% of the study area. The spatial distribution exhibits a distinct pattern, decreasing gradually from east to west and from the coast inland. This study provides a valuable tool for decision-makers to propose targeted adaptation strategies and measures based on the assessment results, thereby better addressing the challenges posed by climate change-induced high-temperature risks and promoting sustainable urban development. Full article

Background: Ovarian cancer is often diagnosed at advanced stages due to the absence of specific early symptoms, resulting in high mortality rates. This study aims to develop a robust and interpretable machine learning (ML) model for the early detection of ovarian cancer, enhancing its transparency through the use of the Contrastive Explanation Method (CEM), an advanced technique within the field of explainable artificial intelligence (XAI). Methods: An open-access dataset of 349 patients with ovarian cancer or benign ovarian tumors was used. To improve reliability, the dataset was augmented via bootstrap resampling. A three-layer deep neural network was trained on normalized demographic, biochemical, and tumor marker features. Model performance was measured using accuracy, sensitivity, specificity, F1-score, and the Matthews correlation coefficient. CEM was used to explain the model’s classification results, showing which factors push the model toward “Cancer” or “No Cancer” decisions. Results: The model achieved high diagnostic performance, with an accuracy of 95%, sensitivity of 96.2%, and specificity of 93.5%. CEM analysis identified lymphocyte count (CEM value: 1.36), red blood cell count (1.18), plateletcrit (0.036), and platelet count (0.384) as the strongest positive contributors to the “Cancer” classification, with lymphocyte count demonstrating the highest positive relevance, underscoring its critical role in cancer detection. In contrast, age (change from −0.13 to +0.23) and HE4 (change from −0.43 to −0.05) emerged as key factors in reversing classifications, requiring substantial hypothetical increases to shift classification toward the “No Cancer” class. Among benign cases, a significant reduction in RBC count emerged as the strongest determinant driving a shift in classification. Overall, CEM effectively explained both the primary features influencing the model’s classification results and the magnitude of changes necessary to alter its outputs. Conclusions: Using CEM with ML allowed clear and trustworthy detection of early ovarian cancer. This combined approach shows the promise of XAI in assisting clinicians in making decisions in gynecologic oncology. Full article

The optimal scheduling of vehicles at signal-free intersections under the connected and automated vehicle (CAV) environment has become a research hotspot in the intelligent transportation field. However, existing studies often oversimplify the intersection’s conflict area and fail to adequately address the spatiotemporal sparsity of conflict points, with little attention given to bus priority requirements. To address these gaps, this paper first establishes an intersection coordinate system and constructs a conflict area analysis model based on the coordinates of key conflict points and vehicle trajectories. Subsequently, an optimal scheduling model for automated vehicles at signal-free intersections with bus priority is developed, which considers the set of vehicles influencing decisions within a time window and uses vehicle entry times and lateral lane changes as decision variables. To enhance computational speed while preserving convergence accuracy, a search space reduction method based on available gaps for conflict point traversal constraints is designed. The model is then solved using an improved double-layer multi-population particle swarm optimization (PSO) algorithm. Simulation results, compared against traditional signal control, rule-driven signal-free, and dynamic-optimization-based signal-free algorithms demonstrate that the proposed method achieves a favorable balance between computational cost and efficiency. It significantly reduces the average vehicle delay. Moreover, incorporating bus priority reduces the average per capita delay by 18.95% compared to the non-priority scenario, effectively proving the validity of the proposed method. Full article

Accurate forecasting of potential collision risk in dense traffic is addressed by a framework grounded in multi-vehicle trajectory prediction. A spatio-temporal fusion architecture, STGAT-EDGRU, is proposed. A Transformer encoder learns temporal motion patterns from each vehicle’s history; a boundary-aware graph (GAT) attention network models inter-vehicle interactions; and a Gated Multimodal Unit (GMU) adaptively fuses the temporal and spatial streams. Future positions are parameterized as bivariate Gaussians and decoded by a two-layer GRU. Using probabilistic trajectory forecasts for the main vehicle and its surrounding vehicles, collision probability and collision intensity are computed at each prediction instant and integrated via a weighted scheme into a Collision Risk Index (CRI) that characterizes risk over the entire horizon. On HighD, for 3–5 s horizons, average RMSE reductions of 0.02 m, 0.12 m, and 0.26 m over a GAT-Transformer baseline are achieved. In high-risk lane-change scenarios, CRI issues warnings 0.4–0.6 s earlier and maintains a stable response across the high-risk interval. These findings substantiate improved long-horizon accuracy together with earlier and more reliable risk perception, and indicate practical utility for lane-change assistance, where CRI can trigger early deceleration or abort decisions, and for risk-aware motion planning in intelligent driving. Full article

Climate change threatens the sustainability of historic buildings with increasing extreme weather events, making energy resilience critical. However, studies on energy resilience often lack forward-looking, holistic approaches. This study aims to develop a conceptual framework that includes how Artificial Intelligence (AI) technologies can support energy resilience in historical buildings with data-driven prediction and analysis to increase energy resilience against climate change. This study applied a methodology with four-stage qualitative research techniques, including a systematic literature review (PRISMA method), content analysis, AI integration, and conceptual framework development processes, in the intersections of historical building, energy resilience, and climate change. The findings reveal a significant research gap in the predictive analysis of the resilience of historic buildings and the integration of AI-based tools in the context of climate change. The proposed framework outlines a multi-layered system that includes data collection, performance analysis, scenario-based prediction, and AI-assisted decision-making, aiming to enhance the resilience of the building (including building envelope, thermal, and lifecycle analysis). Consequently, this study provides a theoretical and methodological perspective and proposes a scientifically based and applicable roadmap. It also highlights the potential of AI as a bridge between energy resilience and historical buildings in the face of a rapidly changing climate. Full article

For the decision-making challenge of ensuring vehicle lane-changing safety, this study proposes a context-dependent causal inference-based model for safe lane changes. Emphasizing multi-vehicle interactions within dynamic traffic scenarios, we construct a three-layer decision-making framework that relies on real-time data collection of speed, acceleration, and spacing information from both the target vehicle and adjacent-lane vehicles. The framework consists of (1) a context-aware layer that extracts standardized dynamic features; (2) an attention mechanism layer that dynamically assigns weights to critical risk factors; and (3) a counterfactual causal reasoning layer where lane-changing risks are quantified through virtual interventions, with multi-objective safety strategies optimized via particle swarm algorithms. The simulation results indicate significant enhancements in high-density traffic conditions. When compared to traditional safety distance models and built-in models from simulation software (SUMO v1.18.0), the proposed model achieves reductions in average conflict counts by 63.0% (from 12.7 to 4.7 instances) and by 37.3% (from 7.5 to 4.7 instances), respectively. Additionally, lane-changing durations are reduced by 10.9% (from 5.5 to 4.9 s) and by 31.9% (from 7.2 to 4.9 s), while fluctuations in risk values decrease by 53.3% (from 0.75 to 0.35) and by 36.4% (from 0.55 to 0.35), respectively. The experimental validation confirms that the integration of dynamic safety distance computation with causal reasoning significantly enhances decision-making robustness in complex scenarios through coordinated risk quantification and multi-objective optimization Full article

This paper introduces a robust and efficient Localized Spline-based Path-Planning (LSPP) algorithm designed to enhance autonomous vehicle navigation on highways. The LSPP approach prioritizes smooth maneuvering, obstacle avoidance, passenger comfort, and adherence to road constraints, including lane boundaries, through optimized trajectory generation using quintic spline functions and a dynamic speed profile. Leveraging real-time data from the vehicle’s sensor fusion module, the LSPP algorithm accurately interprets the positions of surrounding vehicles and obstacles, creating a safe, dynamically feasible path that is relayed to the Model Predictive Control (MPC) track-following module for precise execution. The theoretical distinction of LSPP lies in its modular integration of: (1) a finite state machine (FSM)-based decision-making layer that selects maneuver-specific goal states (e.g., keep lane, change lane left/right); (2) quintic spline optimization to generate smooth, jerk-minimized, and kinematically consistent trajectories; (3) a multi-objective cost evaluation framework that ranks competing paths according to safety, comfort, and efficiency; and (4) a closed-loop MPC controller to ensure real-time trajectory execution with robustness. Extensive simulations conducted in diverse highway scenarios and traffic conditions demonstrate LSPP’s effectiveness in delivering smooth, safe, and computationally efficient trajectories. Results show consistent improvements in lane-keeping accuracy, collision avoidance, enhanced materials wear performance, and planning responsiveness compared to traditional path-planning methods. These findings confirm LSPP’s potential as a practical and high-performance solution for autonomous highway driving. Full article

Continuous and accurate biomass measurement is a critical enabler for control, decision making, and optimization in modern plant production systems. It supports the development of plant growth models for advanced control strategies like model predictive control, and enables responsive, data-driven, and plant state-dependent cultivation. Traditional biomass measurement methods, such as destructive sampling, are time-consuming and unsuitable for high-frequency monitoring. In contrast, image-based estimation using computer vision and deep learning requires frequent retraining and is sensitive to changes in lighting or plant morphology. This work introduces a low-cost, load-cell-based biomass monitoring system tailored for vertical farming applications. The system operates at the level of individual growing trays, offering a valuable middle ground between impractical plant-level sensing and overly coarse rack-level measurements. Tray-level data allow localized control actions, such as adjusting light spectrum and intensity per tray, thereby enhancing the utility of controllable LED systems. This granularity supports layer-specific optimization and anomaly detection, which are not feasible with rack-level feedback. The biomass sensor is easily scalable and can be retrofitted, addressing common challenges such as mechanical noise and thermal drift. It offers a practical and robust solution for biomass monitoring in dynamic, growing environments, enabling finer control and smarter decision making in both commercial and research-oriented vertical farming systems. The developed sensor was tested and validated against manual harvest data, demonstrating high agreement with actual plant biomass and confirming its suitability for integration into vertical farming systems. Full article

Landslides are among the most hazardous natural phenomena affecting Greater Abidjan, causing significant economic and social damage. Strategic planning supported by geographic information systems (GIS) can help mitigate potential losses and enhance disaster resilience. This study evaluates landslide susceptibility using logistic regression and frequency ratio models. The analysis is based on a dataset comprising 54 mapped landslide scarps collected from June 2015 to July 2023, along with 16 thematic predictor variables, including altitude, slope, aspect, profile curvature, plan curvature, drainage area, distance to the drainage network, normalized difference vegetation index (NDVI), and an urban-related layer. A high-resolution (5-m) digital elevation model (DEM), derived from multiple data sources, supports the spatial analysis. The landslide inventory was randomly divided into two subsets: 80% for model calibration and 20% for validation. After optimization and statistical testing, the selected thematic layers were integrated to produce a susceptibility map. The results indicate that 6.3% (0.7 km²) of the study area is classified as very highly susceptible. The proportion of the sample (61.2%) in this class had a frequency ratio estimated to be 20.2. Among the predictive indicators, altitude, slope, SE, S, NW, and NDVI were found to have a positive impact on landslide occurrence. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), demonstrating strong predictive capability. These findings can support informed land-use planning and risk reduction strategies in urban areas. Furthermore, the prediction model should be communicated to and understood by local authorities to facilitate disaster management. The cost function was adopted as a novel approach to delineate hazardous zones. Considering the landslide inventory period, the increasing hazard due to climate change, and the intensification of human activities, a reasoned choice of sample size was made. This informed decision enabled the production of an updated prediction map. Optimal thresholds were then derived to classify areas into high- and low-susceptibility categories. The prediction map will be useful to planners in helping them make decisions and implement protective measures. Full article

Semantic change detection technology based on remote sensing data holds significant importance for urban and rural planning decisions and the monitoring of ground objects. However, simple convolutional networks are limited by the receptive field, cannot fully capture detailed semantic information, and cannot effectively perceive subtle changes and constrain edge information. Therefore, a dynamic graph reasoning network with layer-by-layer semantic decomposition for semantic change detection in remote sensing data is developed in response to these limitations. This network aims to understand and perceive subtle changes in the semantic content of remote sensing data from the image pixel level. On the one hand, low-level semantic information and cross-scale spatial local feature details are obtained by dividing subspaces and decomposing convolutional layers with significant kernel expansion. Semantic selection aggregation is used to enhance the characterization of global and contextual semantics. Meanwhile, the initial multi-scale local spatial semantics are screened and re-aggregated to improve the characterization of significant features. On the other hand, at the encoding stage, the weight-sharing approach is employed to align the positions of ground objects in the change area and generate more comprehensive encoding information. Meanwhile, the dynamic graph reasoning module is used to decode the encoded semantics layer by layer to investigate the hidden associations between pixels in the neighborhood. In addition, the edge constraint module is used to constrain boundary pixels and reduce semantic ambiguity. The weighted loss function supervises and optimizes each module separately to enable the network to acquire the optimal feature representation. Finally, experimental results on three open-source datasets, such as SECOND, HIUSD, and Landsat-SCD, show that the proposed method achieves good performance, with an SCD score reaching 35.65%, 98.33%, and 67.29%, respectively. Full article

Among marine organisms, baleen whale species like the humpback whale (Megaptera novaeangliae) are a case for which individual-based models are necessary to study population changes because individual trait variabilities predominate over average demographic rates to govern population dynamics. These models require quantification of individual organisms’ dynamics with respect to local conditions, which implies optimal strategy frameworks cannot be used. Instead, to quantify how individuals perform according to the environmental conditions they encounter, we formulated a model linking individual mechanical characteristics of swimming and diving with their aerobic metabolism and behavior. The model simulates the dynamics of swimming and diving for the reported range of whale sizes (1000 to 50,000 kg). Additional processes simulate foraging events including rapid accelerations and water engulfment, which modifies whale shape, weight and drag. Simulations show how the energy cost of swimming at equilibrium increases geometrically with velocity and linearly with mass. The duration and distance covered under apnea vary monotonically with mass but not with velocity; hence, there is a positive mass-dependent optimal velocity that maximizes the distance and duration of apnea. The dive limit was explored with a combination of the physiological state, mechanical force produced and distance to return to surface. This combination is imposed as an inequality constraint on the whale individual. The inequality constraint, transformed as a multi-layer perceptron, which continuously processes information about oxygen, depth and relative velocity, provides the whale individual with autonomous decision-making about whether or not to continue the dive. The results also highlight where missing metabolic information is needed to simulate the dynamics of a population of autonomous individuals at the scale of the Global Ocean. Full article

Urbanization is causing a decrease in agricultural land. This leads to changes in cropping patterns, irrigation water availability, and water allowance. Therefore, change in cropping pattern, irrigation water availability, and water allowance were investigated in the Multan region of Pakistan using remote sensing and GIS techniques. The multi-temporal Landsat images with 30 m resolution were acquired for both Rabi (winter) and Kharif (summer) seasons for the years of 1988, 1999 and 2020. The image processing tasks including layer stacking, sub-setting, land use/land cover (LULC) classification, and accuracy assessment were performed using ERDAS Imagine (2015) software. The LULC maps showed a considerable shift of orchard area to urban settlements and other crops. About 82% of orchard areas have shifted to urban settlements and other crops from 1988 to 2020. The LULC maps for Kharif season indicated that cropped areas for cotton have decreased by 42.5% and the cropped areas for rice have increased by 718% in the last 32 years (1988–2020). During the rabi season, the cropped areas for wheat (Triticum aestivum L.) have increased by 27% from 1988 to 2020. The irrigation water availability and water allowance have increased up to 125 and 110% due to decrease in agricultural land, respectively. The overall average accuracies were found as 87 and 89% for Rabi and Kharif crops, respectively. The LULC mapping technique may be used to develop a decision support system for evaluating the changes in cropping pattern and their impacts on net water availability and water allowances. Full article

Change detection (CD) is essential for Earth observation tasks, as it identifies alterations in specific geographic areas over time. The advancement of deep learning has significantly improved the accuracy of CD. However, encoder–decoder architectures often struggle to effectively capture temporal differences. Encoders may lose critical spatial details, while decoders can introduce bias due to inconsistent receptive fields across layers. To address these limitations, this paper proposes an enhanced symmetric network, termed FDRF (feature difference and receptive field), which incorporates two novel components: the multibranch feature difference extraction (MFDE) module and the adaptive ensemble decision (AED) module. MFDE independently extracts differential features from bitemporal images at each encoder layer, using multiscale fusion to retain image content and improve the quality of feature difference modeling. AED assigns confidence weights to predictions from different decoder layers based on their receptive field sizes and then combines them adaptively to reduce scale-related bias. To validate the effectiveness and robustness of FDRF, experiments were conducted on five public datasets (SYSU, LEVIR-CD, WHU, NJDS, and CLCD), as well as a UAV-based dataset collected from two national coastal nature reserves in Guangxi Beihai, China. The results demonstrate that FDRF consistently outperforms existing methods in accuracy and robustness across diverse scenarios. Full article

Accurate measurement and understanding of the spatiotemporal distribution of soil water content (SWC) are crucial in various environmental and agricultural sectors. The present study implements a novel precision agriculture (PA) approach under sugarbeet field conditions of two moisture-irrigation treatments with two subfactors, clay loam (CL) and clay (C) soils, for geostatistics modeling (seven models’ evaluation) of time domain reflectometry (TDR) multisensor network measurements. Two different sensor calibration methods (M1 and M2) were trialed, as well as the results of laboratory soil analysis for geospatial two-dimensional (2D) imaging for accurate GIS maps of root zone moisture profiles, granular, and hydraulic profiles in multiple soil layers (0–75 cm depth). Modeling results revealed that the best-fitted semi-variogram models for the granular attributes were circular, exponential, pentaspherical, and spherical, while for hydraulic attributes were found to be exponential, circular, and spherical models. The results showed that kriging modeling, spatial and temporal imaging for accurate profile SWC ${\theta v}_{TDR}$ (m³·m⁻³) maps, the exponential model was identified as the most appropriate with TDR sensors using calibration M1, and the exponential and spherical models were the most appropriate when using calibration M2. The resulting PA profile maps depict spatiotemporal soil water variability with very high resolutions at the centimeter scale. The best validation measures of PA profile SWC ${\theta v}_{TDR}$ maps obtained were Nash-Sutcliffe model efficiency NSE = 0.6657, MPE = 0.00013, RMSE = 0.0385, MSPE = −0.0022, RMSSE = 1.6907, ASE = 0.0418, and MSDR = 0.9695. The sensor results using calibration M2 were found to be more valuable in environmental irrigation decision-making for a more accurate and timely decision on actual crop irrigation, with the lowest statistical and geostatistical errors. The best validation measures for accurate profile SWC ${\theta v}_{TDR}$ (m³·m⁻³) maps obtained for clay loam over clay soils. Visualizing the SWC results and their temporal changes via root zone profile geostatistical maps assists farmers and scientists in making informed and timely environmental irrigation decisions, optimizing energy, saving water, increasing water-use efficiency and crop production, reducing costs, and managing water–soil resources sustainably. Full article