Search Results (42)

Background and Objectives: Termination of pregnancy (TOP) for fetal anomalies is a clinically and ethically complex intervention, and epidemiological data from regional Turkish referral centers are limited. This study aims to characterize the distribution of indications, identify predictors of late termination (gestational age ≥ 20 weeks), and analyze hospital resource utilization in a nine-year consecutive case series. Materials and Methods: A retrospective cohort study was conducted. All 112 cases of TOP for confirmed fetal anomalies or fetomaternal indications were included. Predictors of late termination were identified by binary logistic regression with three a priori selected variables (cardiovascular anomaly, CNS anomaly, maternal age; EPV [events-per-variable ratio] = 13.3). Hospital stay was compared between gestational age groups using the Mann–Whitney U test. Results: The mean gestational age at termination was 18.2 ± 4.0 weeks. Amniotic fluid disorders (predominantly anhidramnios, n = 35, 31.3%) were the single most frequent specific diagnosis. Cardiovascular anomaly was the only variable associated with late termination on logistic regression (OR 15.34, 95% CI 1.81–130.34; p = 0.012); CNS anomaly and maternal age were not significant. Cardiovascular anomalies were terminated at significantly later gestational ages than all other indication categories (pairwise Mann–Whitney U; Bonferroni-corrected p < 0.01 for four of five comparisons). Hospital stay was significantly longer in cases terminated at ≥20 weeks (median 3 vs. 2 days; p = 0.001). Conclusions: Amniotic fluid disorders dominate the indication profile at this center, distinguishing it from chromosomally focused Western series. Cardiovascular anomalies were associated with late termination, a finding consistent with the gestational age constraints of fetal echocardiography. These findings have implications for prenatal screening program design and healthcare resource planning. Full article

The proliferation of Internet of Things (IoT) devices operating at the network edge has created unprecedented demand for distributed machine learning capable of functioning under severe resource constraints. Federated learning (FL) has emerged as a promising paradigm for privacy-preserving collaborative model training across distributed nodes; however, its application to energy-constrained edge environments remains insufficiently characterized at the system level, particularly with respect to reproducible evaluation of resource consumption and communication efficiency. In this paper, we present EcoFL (Energy-Conscious Federated Learning), a modular, energy-aware benchmarking and orchestration framework for systematic evaluation of lightweight machine learning models under emulated edge hardware constraints. Rather than proposing a new federated optimization algorithm, EcoFL extends a standard FedAvg-based training pipeline with three principal components: (i) an energy-aware communication scheduler that dynamically adapts aggregation rounds and client participation based on per-node resource availability; (ii) a comprehensive system-level profiling pipeline capturing CPU utilization, RAM consumption, inference latency, communication overhead, and estimated computational energy consumption per training round; and (iii) a reproducible benchmarking methodology enabling fair comparison of centralized, standard federated (FedAvg), and energy-aware federated configurations. We evaluate five lightweight model families—Logistic Regression, Random Forest, XGBoost, Multilayer Perceptron, and Isolation Forest—under emulated Raspberry Pi 4 hardware constraints using an anomaly detection task on synthetic IoT sensor telemetry (50,000 samples, 12 features, Dirichlet non-IID partitioning). Experimental results across five independent seeds show that, within the evaluated benchmark setting, EcoFL reduces estimated federated training energy by 79.9–92.9% (mean 84.4%) relative to standard FedAvg through adaptive round termination (4–7 rounds versus 20 fixed rounds), while showing no statistically significant F1-score degradation for four of the five evaluated model families under the tested seed regime. Notably, EcoFL achieves a higher F1-score than FedAvg for Random Forest (+0.052), which we attribute to reduced overfitting resulting from earlier convergence under non-IID data distributions. The full EcoFL framework is released as open-source software to promote reproducibility in energy-aware federated learning research and to facilitate systematic investigation of the trade-offs between predictive performance, resource utilization, and communication overhead in resource-constrained edge environments. Full article

Agricultural cold chain logistics is characterized by inherent challenges—product perishability, high carbon emissions, and stringent time windows—which are further exacerbated by dynamic disruptions. Existing methods suffer from slow adaptability, unstable multi-objective convergence, and severe cold-start issues. This work falls within the broad scope of biomimetics—the science of emulating nature’s time-tested strategies to solve complex engineering problems—and bio-inspired data-driven methods and their applications in engineering control, optimization, and artificial intelligence. The proposed H-MODRL framework embodies core biomimetic principles: the Genetic Algorithm (GA) mimics Darwinian natural selection and genetic inheritance, the Sparrow Search Algorithm (SSA) abstracts the cooperative foraging and anti-predation behaviors of sparrow populations in nature, and the Arrhenius-based freshness-decay model captures the biochemical kinetics governing perishable biological products. By synergistically integrating these biological evolution principles, swarm intelligence, and deep learning, the framework tackles real-world logistics complexity in a manner directly inspired by living systems. This study presents a well-organized hybrid optimization framework (H-MODRL) that couples a three-stage hybrid evolutionary mechanism, synergistically integrating heuristic warm-start, evolutionary policy guidance, and deep reinforcement learning decision-making. First, an improved genetic algorithm combined with the earliest deadline first strategy constructs a feasible initial population satisfying hard time-window constraints. Second, a large neighborhood search-enhanced chaotic sparrow search algorithm builds a high-quality elite guidance set for policy learning. Third, a physics-based multi-objective proximal policy optimization model embedded with Arrhenius equation-derived freshness-decay kinetics performs online decision-making. Experiments demonstrate that pre-computed all-pairs shortest paths and an O(1) hash-based dynamic-disruption indexing mechanism support fast online replanning. On heterogeneous simulated terrains based on real Chinese geospatial data, H-MODRL outperforms state-of-the-art algorithms across four objectives—logistics cost, carbon emissions, terminal freshness, and delivery time—while exhibiting compact, low-variance performance distributions, thereby validating its engineering robustness and practical value in complex agricultural cold chain environments. Full article

As representatives of early-flowering herbaceous peony types, certain cultivars known as the ‘Coral’ series are highly prized in the global cut flowers market for their unique dynamic color transitions from orange-red (amber) to creamy yellow during the florescence and senescence periods. Despite their strong growth vigor and high commercial value, these cultivars face critical postharvest preservation challenges, most notably rapid petal abscission and short vase life. Previous studies have confirmed that postharvest quality deterioration of these peony cut flowers, including undesired color fading and accelerated senescence of petals, is closely associated with ethylene and ROS accumulation. To address these development impediments, systematic optimization of the entire industrial chain is essential. Proposed countermeasures include preharvest regulation of environmental conditions and cultivation practices to establish a foundation for quality formation, as well as postharvest strategies such as precise harvest timing, anti-ethylene treatments, and full cold-chain logistics. Meanwhile, simplifying the distribution system and optimizing terminal vase preservation techniques are also crucial to maintain postharvest quality. In the long term, promoting sustainable development of the global cut-flower industry will require breeding new germplasm with low ethylene sensitivity from a global perspective, continuously optimizing agronomic practices to overcome year-round supply constraints, and accelerating the application of intelligent technologies such as the Internet of Things (IoT) in full chain quality management. Full article

Withthe widespread deployment of intelligent terminals, mobile payment platforms, and Internet of Things devices, security systems are being progressively transformed from traditional transaction outcome analysis toward an intelligent perception paradigm centered on user behavior, device states, and environmental context. To address the challenges of multimodal data heterogeneity, non-independent and identically distributed data across nodes, and the difficulty of centralized modeling under privacy constraints in distributed scenarios, an artificial intelligence-driven federated multimodal security perception framework, namely FMS-LLM, is proposed. At its core, the framework introduces a Non-IID adaptive federated fusion mechanism that achieves dual-level alignment—structural alignment via parameter-level masks and semantic alignment via feature consistency constraints—to effectively mitigate cross-node distribution discrepancies. Additionally, an LLM-driven semantic enhancement module is developed, utilizing trend-guided token selection and inertia-suppression to map low-level sensing features into high-level risk semantic representations, thereby supporting logical reasoning and explainable decision-making. This framework takes user behavioral sensing data, device state information, environmental context data, and transaction behavior data as inputs, and constructs an integrated security analysis pipeline of “perception–collaboration–reasoning”. Experimental results on the distributed multimodal security perception task demonstrate that the proposed method achieves an Accuracy of 91.62%, a Precision of 91.04%, a Recall of 90.37%, an F1-score of 90.70%, and a ROC-AUC of 94.73%, consistently outperforming baseline methods including Logistic Regression, Random Forest, LSTM, the centralized multimodal deep model, FedAvg, FedProx, and MOON. Under strongly Non-IID conditions, when $\alpha =0.1$, the model still maintains an Accuracy of 88.47% and an F1-score of 87.11%, demonstrating stronger cross-node robustness. The ablation study further indicates that the complete model attains the best classification performance while reducing communication cost to 18.92 MB/Round. These results demonstrate that the proposed method can effectively fuse multi-source sensing information under privacy-preserving conditions and support intelligent security perception tasks with higher accuracy, stronger robustness, and improved interpretability. Full article

Driven by the increasing reliance on maritime transportation, achieving energy efficiency and reducing emissions in automated container terminals (ACTs) has become an urgent priority. The complex structure of a low-carbon terminal, along with the interdependent relationships between ACT container operation system and energy system, presents significant challenges to container terminal management. This issue has not yet been discussed in the existing research literature. To this end, this paper introduces a co-optimization problem of the operation, logistics, and energy (CPOL&E) based on the coupling mechanism between multiple subsystems in the ACTs. The CPOL&E considers energy recovery, where the cranes equipped with super-capacitors can recover energy during container unloading and release the stored energy during container lifting to effectively reduce the total energy consumption. A co-scheduling model is developed to simultaneously minimize the makespan and the terminal energy cost. To solve this multi-objective model, we design a multi-subpopulation differential co-evolution algorithm (MDCEA). In this algorithm, an initialization method based on heuristic rules and population entropy is introduced. The MDCEA employs a differential multi-subpopulation co-evolution strategy combined with an adaptive local search approach to enhance population performance. Through comparative experiments with three representative multi-objective algorithms, the results indicate that the MDCEA is capable of obtaining high-quality solutions and demonstrates competitive performance in solving the CPOL&E under the tested instances. Simulation results of the co-optimization mode demonstrate that the scheduling of terminal equipment significantly affects the temporal distribution of its energy demand, which in turn influences the output of the energy system. Full article

Intermodal container terminals play a critical role in modern logistics systems, where operational efficiency and energy consumption are strongly influenced by container stacking strategies. Inefficient yard organization leads to increased reshuffling operations, which negatively affect handling time and resource utilization. Despite extensive research, the relationship between operational performance and energy consumption remains insufficiently explored under dynamic terminal conditions. This study applies a discrete-event simulation framework to evaluate the impact of alternative container stacking strategies on both operational efficiency and energy consumption. The model represents container arrivals, storage decisions, retrieval processes, and reshuffling operations over a multi-day simulation horizon. Three stacking strategies—FIFO, balanced distribution, and departure-time clustering—are analysed under identical and dynamically evolving conditions using performance indicators related to reshuffling intensity, handling efficiency, and energy consumption. The results show that stacking strategies significantly affect terminal performance, but their effectiveness depends on the structure of container flows. While FIFO achieves the lowest reshuffling intensity and energy consumption under high-load conditions, departure-time clustering improves performance in outbound-dominated scenarios. The findings also reveal a structural discrepancy between operational and energy-related performance, as non-productive movements account for a higher share of operations than of total energy consumption. The study demonstrates that container stacking should be treated as a multi-criteria decision problem, where minimizing reshuffles does not directly correspond to minimizing energy consumption. The proposed simulation-based framework provides a consistent environment for evaluating trade-offs between operational and energy-related performance under controlled dynamic conditions. Full article

Dry ports are becoming increasingly important elements of port–hinterland transport systems, particularly as maritime gateways face rising congestion, infrastructure pressure, and coordination challenges within global supply chains. As international trade expands and logistics networks grow more complex, inland terminals are progressively evolving into integrated intermodal platforms that support more efficient freight distribution between seaports and their hinterlands. This study presents a PRISMA-based systematic review of research on dry port–seaport systems covering the period 1980–2025. Following a structured screening and selection procedure, peer-reviewed publications were identified and analyzed to examine conceptual developments, thematic orientations, geographical scope, and decision-making perspectives within the field. Particular attention is given to the growing relevance of digital transformation, including artificial intelligence and machine learning, in shaping future dry port operations and network design. By synthesizing existing contributions and identifying research gaps, this review provides a consolidated understanding of the evolution of dry port research and outlines key directions for advancing sustainable, resilient, and data-driven port–hinterland systems. Full article

Objective: To evaluate clinical factors associated with termination of pregnancy (TOP) recommendations following prenatal diagnosis of congenital heart disease (CHD) within a multidisciplinary fetal council model. Methods: This retrospective cohort study included 146 fetuses with prenatally diagnosed CHD discussed in a tertiary referral center fetal council between October 2023 and December 2025. The primary outcome was council-issued recommendation for TOP (yes/no). Variables included gestational age (GA) at diagnosis, cardiac severity (Davey scale grouped as low, moderate, high), extracardiac anomalies, fetal growth restriction (FGR), and genetic evaluation/results. Group comparisons were performed using Mann–Whitney U and χ² tests. Independent associations were assessed using binary logistic regression. A subgroup analysis was conducted in isolated CHD cases (no extracardiac structural anomalies). Results: A total of 146 fetuses with prenatal CHD were included in the analysis. TOP was recommended in 71 cases (48.6%). GA at diagnosis did not differ between groups when analyzed continuously; however, categorical GA showed significant differences, with earlier diagnoses more frequent among TOP-recommended cases. Cardiac severity distribution differed significantly between groups. In multivariable analysis, GA at diagnosis and cardiac severity were independently associated with TOP recommendation. Compared with <20 weeks, diagnosis at 20–23+6 weeks (OR 17.96, 95% CI 3.50–92.22) and ≥24 weeks (OR 3.92, 95% CI 1.53–10.06) increased the odds of TOP recommendation. Relative to high severity, moderate (OR 0.23, 95% CI 0.07–0.72) and low severity (OR 0.20, 95% CI 0.08–0.50) were associated with lower odds of TOP recommendation. Extracardiac anomalies, genetic findings, and FGR were not independently associated after adjustment. Similar patterns were observed in isolated CHD cases. Conclusions: In a multidisciplinary prenatal counseling setting, TOP recommendations after prenatal CHD diagnosis were primarily driven by cardiac severity and GA at diagnosis, rather than extracardiac or genetic findings alone. Full article

Background: Liquefied natural gas (LNG) distribution in archipelagic regions involves complex trade-offs between transportation, infrastructure investment, and contractual arrangements. While most optimisation studies focus on seller-managed Delivery Ex-Ship (DES) schemes, limited research addresses buyer-managed Free on Board (FOB) frameworks that extend decision responsibility upstream. Methods: This study develops a two-stage integrated optimisation model for long-term LNG supply chain planning under an FOB contractual scheme with time-dependent deterministic demand. Stage 1 determines hub selection, port clustering, vessel sizing, fleet configuration, and endogenous infrastructure capacities using a genetic algorithm, while Stage 2 optimises cluster-level routing sequences. Robustness is assessed through multiple independent runs and sensitivity analysis. Results: A case study of the Nusa Tenggara region identifies Sumbawa as the optimal hub. The upstream segment consistently selects a 65,000 m³ vessel under terminal service capacity constraints, while downstream clusters are served by 3500 m³ and 10,000 m³ vessels depending on distance and demand aggregation. Infrastructure requirements are derived from peak-demand conditions, and the resulting levelised logistic cost is 4.66 USD/MMBtu. Conclusions: The findings demonstrate that FOB arrangements fundamentally reshape network configuration, fleet segmentation, and infrastructure sizing, providing a robust strategic planning framework for buyer-managed LNG supply chains in archipelagic contexts. Full article

Aim: This study aimed to characterize and compare the composition of central (TCM), effector (TEM), tissue-resident (TRM), and terminally differentiated (TEMRA) memory T cells in mid-luteal endometrium during the window of implantation (WOI) in women with and without a previous miscarriage. Methods: Stromal lymphocytes from endometrial samples (P + 5) were analyzed by multicolor flow cytometry to quantify total, CD4⁺ and CD8⁺ TCM (CD45RA⁻CCR7⁺), TEM (CD45RA⁻CCR7⁻), TRM (CD69⁺), and TEMRA (CD45RA⁺CCR7⁻) subsets. Participants were grouped as having no previous miscarriage (n = 38) or ≥1 previous miscarriage (n = 33), and the relative distribution of these memory subsets was compared between groups. Correlations, PCA and logistic regression were used to assess global memory network organization. Results: Women with prior miscarriage exhibited higher TCM proportions among total and CD8⁺ lymphocytes (p < 0.01), alongside lower CD8⁺ TEM (p = 0.02) and higher CD4⁺ TEM (p = 0.01). TRM showed a mild, non-significant increase (p = 0.18), while TEMRA remained stable. TRM correlated positively with both TCM (r = 0.51) and CD4⁺ TEM (r = 0.40), indicating coordinated organization among memory subsets. Multivariate analyses (PCA and logistic regression) confirmed these trends and identified the TCM/TEM ratio as the most discriminative parameter. Conclusions: Endometrial memory T-cell composition during the WOI differs in women with miscarriage history, characterized by central memory expansion and reduced effector memory proportions, with parallel increases in tissue-resident cells. These changes suggest persistent remodeling of the local immune memory network toward a long-lived, less differentiated phenotype that may influence implantation readiness in subsequent cycles. Full article

The rapid growth of e-commerce and advances in information and communication technologies have placed increasing pressure on last-mile delivery companies to enhance operational productivity. As investments in logistics infrastructure require long-term planning, maximizing the efficiency of existing terminal operations has become a critical priority. This study proposes a mathematical model for inbound truck scheduling that simultaneously minimizes truck waiting times and balances workload across temporary inventory storage located at outbound chutes in cross-docking terminals. The model incorporates a dynamic rescheduling strategy that updates the assignment of inbound trucks in real time, based on the latest terminal conditions. Numerical experiments, based on real operational data, demonstrate that the proposed approach significantly outperforms conventional strategies such as First-In First-Out (FIFO) and Random assignment in terms of both load balancing and truck turnaround efficiency. In particular, the proposed model improves workload balance by approximately 10% and 12% compared to the FIFO and Random strategies, respectively, and it reduces average truck waiting time by 17% and 18%, thereby contributing to more efficient workflow and alleviating bottlenecks. The findings highlight the practical potential of the proposed strategy for improving the responsiveness and efficiency of parcel distribution centers operating under fixed infrastructure constraints. Future research may extend the proposed approach by incorporating realistic operational factors, such as cargo heterogeneity, uncertain arrivals, and terminal shutdowns due to limited chute storage. Full article

Chickpea is a highly weed-prone crop with limited herbicide options and high labor demands, raising the following question: Can fall-planted legume–cereal cover crops (CCs) improve soil properties while reducing herbicide use and manual weeding pressure? To explore this, we evaluated the effect of fall-planted winter rye (WR) alone in 2021 and mixed with hairy vetch (HV) in 2022 and 2023 at Randolph farm in Petersburg, Virginia. The objectives were two-fold: (a) to examine the effect of CCs on soil properties using monthly growth dynamics and biomass harvested from fifteen 0.25 m²-quadrants and (b) to evaluate the efficiency of five termination methods: (1) green manure (GM); (2) GM plus pre-emergence herbicide (GMH); (3) burn (BOH); (4) crimp mulch (CRM); and (5) mow-mulch (MW) in suppressing weeds in chickpea fields. Weed distribution, particularly nutsedge, was patchy and dominant on the eastern side. Growth dynamics followed an exponential growth rate in fall 2022 (R² ≥ 0.994, p < 0.0002) and a three-parameter sigmoidal curve in 2023 (R² ≥ 0.972, p < 0.0047). Biomass averaged 55.8 and 96.9 t/ha for 2022 and 2023, respectively. GMH consistently outperformed GM in weed suppression, though GM was not significantly different from no-till systems by the season’s end. Kabuli-type chickpeas under GMH had significantly higher yields than desi types. Pooled data fitted well to a three-parametric logistic curve, predicting half-time to 50% weed coverage at 35 (MM), 38 (CRM), 40 (BOH), 46 (GM), and 53 (GMH) days. Relapses of CCs were consistent in no-till systems, especially BOH and MW. Although soil properties improved, CCs alone did not significantly suppress weed. Full article

In recent years, as an important alternative to traditional gasoline-powered vehicles, new electric vehicles (NEVs) have gained widespread attention and rapid development globally. In the traditional automotive industry chain, downstream vehicle manufacturers need to master core technologies, such as engines, chassis, and transmissions. In contrast to the traditional automotive industry chain, where downstream vehicle manufacturers must master core technologies, like engines, chassis, and transmissions, the electric vehicle industry chain has evolved in a way that the development of core components is gradually separated from the vehicle manufacturers. Downstream vehicle manufacturers can now outsource key components, such as batteries, electric controls, and motors. Additionally, in terms of sales models, the electric vehicle industry chain can adopt either the traditional 4S dealership model or a direct-sales model. As the research and development of core components are increasingly separated from vehicle manufacturers, the downstream vehicle manufacturers can source components, like batteries, electric controls, and motors, externally. At the same time, they can choose to use either the traditional 4S dealership model or the direct-sales model. The underlying mechanisms and channel selection in this context require further exploration. Based on this, a mathematical model is established by incorporating terminal marketing input, product competitiveness, and after-sales service levels from the literature to solve for the optimal pricing under centralized and decentralized pricing strategies. Using numerical examples, the pricing and profit performance under different market structures are analyzed to systematically examine the impact of the electric vehicle supply chain on business operations, as well as the changes in various elements across different channels. We will focus on how after-sales services (including the spare part supply) influence the pricing strategy and profit distribution in the supply chain, aiming to provide insights into advanced manufacturing system management for manufacturing enterprises and improve the efficiency of intelligent logistics management. The research indicates that (1) The direct-sales model helps to improve the terminal marketing input, after-sales service quality, and product competitiveness for supply chain stakeholders; (2) It is noteworthy that the manufacturer’s direct-sales model also significantly contributes to lowering prices, highlighting that the direct-sales model has substantial impacts on both supply chain stakeholders and, importantly, consumers. Full article

In recent years, routing problems have attracted significant attention in the fields of operations research and computer science due to their fundamental importance in logistics and transportation. However, most existing learning-based methods employ simplistic context embeddings to represent the routing environment, which constrains their capacity to capture real-time visitation dynamics. To address this limitation, we propose a deep reinforcement learning-based decision-making framework (DRL-DM) built upon an encoder–decoder architecture. The encoder incorporates a batch normalization fronting mechanism and a gate-like threshold block to enhance the quality of node embeddings and improve convergence speed. The decoder constructs a dynamic-aware context embedding that integrates relational information among visited and unvisited nodes, along with the start and terminal locations, thereby enabling effective tracking of real-time state transitions and graph structure variations. Furthermore, the proposed approach exploits the intrinsic symmetry and circularity of routing solutions and adopts an actor–critic training paradigm with multiple parallel trajectories to improve exploration of the solution space. Comprehensive experiments conducted on both synthetic and real-world datasets demonstrate that DRL-DM consistently outperforms heuristic and learning-based baselines, achieving up to an 8.75% reduction in tour length. Moreover, the proposed method exhibits strong generalization capabilities, effectively scaling to larger problem instances and diverse node distributions, thereby highlighting its potential for solving complex, real-life routing tasks. Full article