Background: Diabetes mellitus affects 463 million people worldwide and necessitates continuous blood glucose monitoring. Current glucose prediction systems often lack efficiency, and real-time prediction is essential for timely clinical intervention. Objectives: This study aims to develop and validate a novel Convolutional Recurrent Neural Network (CRNN) enhanced with bio-inspired algorithms to improve blood glucose prediction and enable real-time detection of hypoglycemia and hyperglycemia. Materials and methods: The proposed framework employs a CRNN architecture that combines Convolutional Neural Networks (CNNs) for feature extraction with Long Short-Term Memory (LSTM) layers for temporal sequence learning. The model was trained and evaluated using the HUPA-UCM diabetes dataset. Additionally, the study benchmarks the proposed model against 19 traditional Machine Learning (ML) algorithms and compares it with state-of-the-art methods from the literature. Results: The proposed approach demonstrates superior predictive capability, consistently delivering promising results across multiple evaluation frameworks. The model achieves clinically acceptable prediction intervals, confirming its effectiveness in enhancing the accuracy and reliability of blood glucose prediction for diabetes management. Conclusion: The findings demonstrate that the proposed CRNN model, enhanced with bio-inspired algorithms, provides an effective and reliable solution for real-time blood glucose prediction. By outperforming conventional ML methods and achieving clinically acceptable accuracy levels, the model shows strong potential for integration into intelligent diabetes management systems to support timely clinical decisions and improve patient outcomes.

Short communication: Amitraz treatments do not increase the frequency of mutations in the β2-adrenergic octopamine receptor in Varroa destructor: a field study in Central Spain.

Sustainable agriculture requires maintaining soil health, yet conventional management (CM) practices may not protect soils from stresses such as compaction. This study compared microbial resilience to compaction in two soils collected from sugarcane farms under improved management (IM: minimum tillage, cover cropping and stubble retention) and CM (conventional tillage, no cover crop and stubble retention) practices. Samples were placed in 96-well deep-well plates and compacted using a bespoke device to achieve bulk densities of 0.9 (control), 1.1 (low), and 1.2 g cm⁻³ (moderate). Microbial resistance was assessed 14 days after compaction, and resilience 14 days after stress relief. Under low and moderate compaction, IM soils showed 49.5 % and 45.7 % higher CO₂ emission resistance indices (i.e., the ability of soil to maintain microbial respiration under compaction stress) than CM, indicating greater stability. Microbial biomass carbon and nitrogen were 56.2 % and 47.9 % higher in IM soils under low compaction, compared to CM. Soil microbial metabolic quotient ( q CO₂) was similar across compaction levels within each system, but was 19.5 %–36.3 % lower in IM soils than CM at equivalent compaction, indicating lower microbial stress under IM. Fourteen days after stress relief, q CO₂ in moderately compacted CM soil increased by 41.1 % and 25.0 % compared to control and low compaction. In contrast, IM soil under moderate compaction had 40.6 % lower q CO₂ than CM. The CM showed no effects of compaction on hot water extractable organic carbon content, while compaction of IM showed a 13 % decline compared to its control. Hot water extractable total nitrogen did not vary with compaction within the management systems but was 12 %–15 % higher in IM than CM under the same compaction during the resistance phase. Total mineral nitrogen was unaffected by compaction treatments under each system but was 11 %–13 % higher in IM than CM during resistance phase. These findings highlight the potential of improved management practices to sustain soil health and resilience under compaction stress. • A novel test used to assess microbial functional resilience to compaction stress. • Improved management had higher respiration resistance than conventional management. • Improved management reduced microbial metabolic quotient under compaction stress. • Improved management enhanced microbial functional stability and nutrient retention.

From ‘everywhere’ to ‘anywhere’: integrating local observing with global systems for coastal management

Diabetic ulcers remain a significant clinical concern because of their poor healing, persistent inflammation, and increased infection risk. Chitosan and its derivatives have emerged as promising biomaterials for ulcer management due to their biocompatibility, biodegradability, mucoadhesiveness, pH responsiveness, and intrinsic properties, including hemostatic effects, antibacterial activity, anti-inflammatory properties, antioxidant capabilities, and self-healing capabilities. This review provides a comprehensive and current overview of chitosan-based approaches for diabetic ulcer management, focusing on both conventional and advanced biomaterials, including nanoparticles, hydrogels, nanofibers, microneedles, sponges, and 3D scaffolds. Special attention is given to stimuli-responsive chitosan systems designed to adjust drug release in response to specific cues, such as pH, reactive oxygen species, temperature, enzymes, or glucose levels. To underscore the translation potential, the review compiles commercialized products, recent patents, and clinical trials related to the management of diabetic ulcers based on chitosan formulations. Finally, the integration of AI-driven design with smart chitosan systems is proposed as a future direction towards personalized and predictive diabetic ulcer management.

Application of recommender systems in supply chain management: A state-of-the-art review

Design research on LLM-based intelligent information systems for radiation equipment safety management

Aquatic foods are essential sources of protein and micronutrients and play a critical role in global nutrition, trade, and livelihoods. However, their safety and sustainability are frequently compromised by microbial contamination and biofilm formation during farming, processing, storage, and retail. Biofilms persist on moist surfaces, resist conventional cleaning practices, and contribute to spoilage, cross-contamination, and economic loss. This article reviews emerging applications of artificial intelligence and Industry 4.0 technologies for biofilm prevention and control in aquaculture and seafood systems. Particular emphasis is placed on the use of continuous water quality sensing, imaging platforms for early detection and cleaning verification, genomic and omics tools for microbial trait-level insight, and digital twin frameworks for virtual simulation of sanitation strategies. Recent advances demonstrate that sensor telemetry can predict biofilm-favorable conditions, imaging can verify removal in real time, and genomic data can identify persistence traits and tolerance mechanisms. When integrated, these approaches enable facility-specific digital twins that anticipate surface-specific risks and recommend optimized interventions before implementation. The convergence of AI, sensor networks, imaging, and omics offers a shift from reactive to proactive biofilm management in aquatic food systems. Positioned within the transition to Industry 5.0, these innovations support earlier detection, targeted interventions, and measurable improvements in food safety, quality, sustainability, and resilience, while aligning production systems with human-centric goals.

Ethics label for digital systems to promote transparency and user awareness

African swine fever (ASF) poses an ongoing threat to pig production and wild boar across Europe. Controlling ASF in wild boar builds a complex system with many stakeholders involved. Participatory modelling complements traditional risk assessments by identifying leverage points, revealing practical barriers, and supporting adaptive, context-sensitive management in complex animal health systems. To strengthen ASF preparedness in Switzerland, we applied participatory modelling by engaging stakeholders from two regions in the co-development of semi-quantitative system models for the anticipated control of ASF in wild boar. Through a structured series of workshops, participants collaboratively identified key control measures, their leveraging and hindering factors and potential consequences within and beyond the ASF response system. Within the model, control measures were prioritised based on their expected impact over time within the system. Across regions, the stakeholder-informed model consistently prioritized measures related to coordination, operational communication amongst stakeholders and risk communication to the public, and carcass search and disposal. Many of the top-ranked measures came from thematic areas not covered in that-time existing technical guidelines, reflecting the added value of participatory approaches. Across regions, influencing factors such as coordination structures, legal frameworks, and resource availability, including trained personnel and carcass search dogs, were identified as critical enablers for effective ASF control. The analysis revealed a growing emphasis on clear and centralised government leadership. Beyond model outputs, the participatory modelling process fostered trust, strengthened cross-sectoral networks, and enhanced co-construction of knowledge. These findings highlight the value of participatory approaches for embedding stakeholder expertise into disease control planning, leading to shared ownership of ASF preparedness strategies.

Long-term forecasting of water quality and algal dynamics in riverine systems using advanced physicochemical-informed machine learning models.

Enhancing water management and natural resource use in tropical grassland systems under semi-arid conditions: Dual Kc approach

Three key directions toward the development of a holistic crop modelling framework

Historically, for-profit organizations participate in electricity markets to benefit their shareholders. But as communities increasingly gain the ability to own and operate distributed energy resources (DER), it has become possible to reconsider how the benefits of aggregation are distributed, and shift them towards the community they are located in. This paper proposes the Altruistic Aggregator (AA) framework to maximize community benefit as an alternative to a traditional profit-maximizing aggregator. The AA framework optimally participates in the wholesale electricity and regulation reserves markets and then allocates credits to improve energy-related outcomes across a community. The AA framework is found to be significantly more effective at distributing the benefits of aggregation among households than a traditional profit-maximizing aggregator and operates at greater economic efficiency than a purely equity-seeking aggregation model, based on the Rawls criterion. In a 500-household case study, the proposed model reduced the Gini coefficient of energy burden from 0.444 to as low as 0.135, while maintaining the shortest payback period and lowest electricity rate among models tested. These results highlight the potential for altruism to guide the grid transformation to benefit communities.

In response to large-scale health crises, emergency healthcare infrastructure projects (EHIPs) have been implemented to provide healthcare services rapidly. However, few studies have systematically explored the success of these projects. This study aims to assess the success of EHIPs throughout the whole life cycle using a cloud matter-element model (CMEM). Considering multiple performance indicators that measure the success of EHIPs, a hierarchy model of the project success index is proposed through eight rounds of Delphi surveys. This model comprises 10 key performance indicators (KPIs) that are most appropriate for measuring the success of EHIPs and 20 quantitative metrics aligned with these KPIs across the life cycle. Based on the hierarchy model, the CMEM is applied to evaluate the success of EHIPs, which effectively addresses the fuzziness and randomness inherent in the Delphi process. Two cases were then utilized for verifying the practicality of the proposed project success assessment model. The findings demonstrate that the CMEM-based assessment offers not only significant flexibility but also high accuracy, robustness, and scalability. The CMEM-based assessment seamlessly integrates numerous incompatible indexes and their characteristic values, without being limited by the number of evaluation indexes. Further, sensitivity analyses are conducted to test the robustness of the assessment model and its results. These analyses confirmed that the CMEM is a robust, reliable, and adaptable tool for assessing the success of EHIPs. Theoretically, this study should set an exemplar of aggregating the performance measurement with emergency management and healthcare systems, facilitating an efficient response to major crises. It also provides a novel approach to evaluating multiple-criteria decision-making problems. Practically, this study should enable project management teams to identify and improve weak areas, ensuring continuous success in the EHIP-related domain.

Malignant temporal bone tumors (1941-2025): A bibliometric analysis of publication trends, key contributors, and thematic evolution.

Towards extreme application scenarios: perspectives on artificial intelligence-driven smart energy management systems

Accurate estimation of the State of Charge (SOC) is essential for enhancing the efficiency and reliability of Battery Management Systems (BMS) in Internet of Things (IoT) applications. This study introduces the Pattern-Aware Transformer Model (PATM), an interpretable framework for SOC prediction in Float-Nominal (FN), Constant-Current (CC), and Energy Release (ER) scenarios. PATM extends the standard Transformer architecture by incorporating a pattern embedding mechanism that explicitly encodes operating conditions and directs adaptive attention allocation. A feature engineering pipeline that combines mutual information (MI) ranking and principal component analysis (PCA) reduces dimensionality while preserving physically relevant variables. On real-world data, PATM achieves an RMSE of 2.08 × 10 −3 and an R 2 of 0.9998, outperforming the Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) baselines. Compared with single-scenario CC modeling, multi-scenario learning reduces RMSE, MAE, MSE, and MAPE by 54.9%, 80.1%, 79.6%, and 75.9%, respectively. Ablation studies further demonstrate that removing the embedding module increases RMSE by 2.4%, MAE by 17.8%, and MSE by 4.9%, while leaving R 2 nearly unchanged. This indicates that the embedding mechanism enhances cross-scenario robustness and error stability. SHapley Additive exPlanations (SHAP) analysis and attention visualizations reveal the model’s dependence on physically relevant factors, including temperature gradients, voltage fluctuations, and internal resistance. • This study combines MI ranking and PCA for efficient feature selection, preserving interpretability while reducing redundancy. • This study introduces a multi-scenario approach that integrates data from diverse operating phases, enhancing adaptability and accuracy. • This study introduces the Pattern-Aware Transformer Model (PATM), and experimental results show that it outperforms LSTM and GRU across multiple metrics, demonstrating its accuracy and robustness.

Although pregnant cows can be assigned an expected calving date, such forecasts remain imprecise due to individual physical and hormonal changes affecting cows' behaviour. Proper management during calving is crucial for the health of cows and calves. This study aimed to predict calving time based on behavioural symptoms in cows of two breeds: 38 Polish Holstein-Friesian (PHF) and 14 Brown Swiss cows from a single farm. Using CowManager sensors, the behaviour of the dairy cows was monitored 24h a day, with data from 3D accelerometers classified into specific activities (eating, ruminating, inactive, active, highly active). As the study was conducted on a single farm, the generalisability of the results to other herd management systems or environmental conditions may be limited. A preliminary graphical analysis identified changes in behaviour in the last hours before calving. Statistical analysis included the bootstrap method, logistic regression, and analysis of change points in the time series (separately for each cow and trait, for moving averages covering 6h). One-day periods were considered in the analysis, starting from 168h before calving. The daily period was shifted by 1h until 6h before calving. The applied methodology showed satisfactory effectiveness (recall of 81.03% and precision of 66.75% for PHF cows). Differences in precalving behaviour between breeds were observed. These findings indicate that sensor-based monitoring of behaviour can provide timely predictions of calving and highlight breed-specific behavioural differences, supporting farm management and animal welfare.

Three-dimensional natural convection and heat transfer between cold plates induced by cross-shaped cylinder: Effects of vertical position