Future Technology

AI in strategic management and organizational agility of SMEs: leadership, policy environment, and adaptive capability

2025-11-26T06:16:02+00:00

This study investigates how artificial intelligence strategic capabilities, transformational leadership, and policy environments collectively influence organizational agility in small and medium-sized enterprises through dynamic capability mechanisms. Employing a mixed-methods design, the research analyzes survey data from 300 SMEs across manufacturing, service, and technology sectors, complemented by qualitative case studies. Structural equation modeling reveals that AI strategic capabilities constitute the strongest predictor of organizational agility (β=0.42, p<0.001), with digital dynamic capabilities mediating 67% of this total effect. Technology-management fit emerges as a critical boundary condition, amplifying AI effectiveness by 123% under high alignment scenarios (β=0.58 versus β=0.26 in low alignment contexts). Transformational leadership exhibits dual mechanisms through direct positive effects on agility (β=0.28, p<0.001) and moderating influences on AI-agility relationships (β=0.21, p<0.01). Notably, AI capabilities demonstrate buffering properties against policy environment uncertainty (β=0.12, p<0.05), transforming institutional constraints into manageable strategic variables. Machine learning analyses reveal nonlinear effects with diminishing returns beyond the 75th percentile of AI adoption. The structural model explains substantial variance in organizational agility (R²=0.64) and firm performance (R²=0.52). These findings extend dynamic capability theory to digital contexts, reconceptualize AI as a strategic capability rather than an operational tool, and illuminate digital leadership dimensions, offering evidence-based guidance for SME managers, technology vendors, and policymakers navigating digital transformation challenges.

Edge-AI microservice orchestration for private, real-time generative FinTech applications

2025-11-25T10:20:32+00:00

The financial services industry faces mounting pressures to deliver real-time, personalized services while safeguarding sensitive user data under tight regulatory environments. Yet, prevailing AI systems in FinTech remain largely cloud dependent, which introduces latency bottlenecks, privacy exposure, and compliance risk. Meanwhile, industry analyses suggest that Edge AI is rapidly becoming a foundational shift, with predictions that 60% of AI deployments will run partially on device by 2029. However, existing edge AI research often focuses on inference optimization, not full-stack orchestration of financial microservices, and therefore, lacks the integrated, decision-oriented intelligence that is required to operate wholly on the device. In this work, we present an architecture for on-device microservice orchestration of generative AI tailored for FinTech use cases. Our system modularizes AI tasks, such as local LLM inference, fraud detection, biometric authentication, and credit scoring, into services coordinated via lightweight orchestrators (e.g. WASMEdge, Open Horizon). Unlike prior approaches, our system coordinates these services using lightweight WebAssembly-based runtimes, enabling secure, isolated, and efficient execution even on resource-constrained devices. Sensitive data, such as transaction history and biometric templates, remains strictly local, with optional federated synchronization for global fraud pattern sharing. With quantized LLMs, we attain inference latency under 90ms, while local anomaly detection achieves 72% accuracy in simulated financial fraud scenarios. The architecture integrates modular microservices, privacy-first orchestration, and a hybrid federated intelligence layer and is among the first to present a decentralized, compliant, and performance-sensitive AI infrastructure for the FinTech of reality.

AI-enabled ESG consultancy capability building in small audit firms: a strategic transformation pathway analysis

2025-11-26T10:52:17+00:00

This study investigates how small audit firms develop artificial intelligence-enabled environmental, social, and governance (ESG) consulting capabilities through strategic transformation pathways. Drawing upon dynamic capabilities theory and employing a mixed-methods research design, the research integrates qualitative interviews with 28 practitioners across 18 small audit firms and quantitative surveys from 156 respondents representing 87 organizations, supplemented by six in-depth case studies. The empirical analysis reveals substantial capability heterogeneity, with only 12.2% of firms deploying advanced AI applications and merely 11.5% achieving established ESG service maturity levels. A significant positive correlation between AI Intensity Index and Capability Maturity Index (r=0.68, p<0.01) demonstrates mutually reinforcing dynamics between technological adoption and domain expertise development. Cluster analysis identifies three transformation pathway archetypes—Technology-Led, Knowledge-Led, and Balanced—with Knowledge-Led approaches achieving marginally higher success rates (72.3%) compared with Technology-Led pathways (67.5%). The research develops a four-stage transformation pathway model spanning 30-36 months, delineating critical activities, resource requirements, and success indicators across foundation, development, integration, and optimization phases. AI-enabled resource optimization yields average efficiency improvements of 46.2% across operational functions, with report generation achieving 61.7% gains. The findings extend dynamic capabilities theory to resource-constrained professional services contexts and provide evidence-based guidance for practitioners navigating digital transformation in expanding ESG advisory markets.

AI-augmented customer-owned channels and relationship value: a multi-mechanism model of engagement, trust, and loyalty

2025-11-27T06:26:27+00:00

This study explains the processes through which a relationship value in customer-owned channels is generated through AI-powered functionalities by focusing on the integrated psychological and behavioral processes. The basis of the work is large-scale text analysis of 18,456 market reviews of applications among 15-20 top customers in the areas of retail, finance, and life services, examining the entire value chain by semantic BERT analysis, hierarchical regression, and structural equation modeling with 5,000 bootstraps. Analysis of the results reveals the presence of function–dimension matching phenomena. These include intelligent recommendation, which has the maximum cognitive engagement (β = 0.41), chatbots, which have the peak affective engagement (β = 0.45), and predictive services, which are predominant in behavioral engagement with a β of 0.36. Customer-owned platforms produce 37.2% more overall effects than third-party platforms. The role of the overall chain mediating function of AI functions for relationship value creation is also supported by the study. The role that loyalty plays between trust and value is amplified by ownership of the channel. The results above present the objective assessment of the 37% boost that the ROI (Return on Investment) has on customer-owned communication channels. It also explains the role that development considerations play.

AI-powered organizational transformation: the role of digital mindset, change management, and cross-cultural leadership in shaping future business strategies

2025-11-25T11:05:47+00:00

This study explores how artificial intelligence reshapes business strategies through synergistic effects between digital thinking, change management, and cross-cultural leadership in organizational transformation processes. Based on multi-source public data from 450 global enterprises across technology, manufacturing, finance, and retail sectors, this research integrates structural equation modeling, in-depth case analysis of 20 extreme cases, and machine learning prediction methods to construct and validate an “AI-Driven Strategic Triple Helix Evolution Framework” through seven interrelated hypotheses. Empirical findings confirm that organizational transformation plays the role of a core mediating hub (R²=0.64), connecting AI capabilities to strategic reconstruction, while the interaction with the three elements of synergy adds an additional 11% of explanatory power to it (ΔR²=0.11, P<0.001). Six strategic paths are differentiated in this research: AI-native (12%), platform transformation (23%), ecosystem orchestration (18%), niche specialization (21%), hybrid innovation (17%), and conservative following (9%), with significant cultural context dependence. Cross-cultural leadership shows the greatest moderating effect on high power distance cultures (β=0.38). The framework goes beyond the traditional technology-organization-environment models in unfolding dynamic co-evolution mechanisms among technological capabilities, cognitive reconstruction, and cultural adaptation. Machine learning models further predict 70% of enterprises participating in ecosystem strategies by 2030, and a digital mindset contributes 34.2% to strategic innovation prediction.

A Feistel-based modification of the PRESENT lightweight block cipher

2025-11-22T16:30:28+00:00

The PRESENT block cipher is commonly used in constrained devices, such as IoT nodes, low-power wireless sensors, and RFID tags, due to its low power consumption, small hardware footprint, and acceptable security margin. However, PRESENT’s substitution layer depends on a fixed S-box, which can reduce resistance against specific analytical attacks (linear and differential analysis). In the proposed version, the static S-box is replaced with a dynamic, key-dependent S-box created by rounds of a lightweight Feistel network. The modified variant eliminates repetitive patterns in the substitution layer. In this study, we evaluate the security strength of the proposed method using differential analysis, linear analysis, avalanche rate, algebraic interpolation attacks, and integral/square attacks. Additionally, we measure the time and hardware complexity. The results demonstrate that the introduced version achieves an average avalanche value of 30.92 flipped bits with a standard deviation of 4.7, indicating strong diffusion with increased variability compared to PRESENT, while maintaining moderate computational and hardware complexity. The improved security of the introduced scheme is attained at the expense of a modest increase in hardware area, consistent with current lightweight cipher design trade-offs.

A descriptive systematic review of contemporary MANET security research: themes, design structures, and reporting rigor

2025-08-21T22:12:20+00:00

Mobile Ad hoc Networks remain vulnerable to a wide range of security threats, yet existing MANET security reviews provide fragmented insight into how recent studies frame security problems, model threats, and report design choices. This descriptive systematic review follows PRISMA guidelines and examines MANET security studies published between January and August 2025. A total of 51 studies were analyzed using a structured coding process covering security themes, attack types, defense techniques, threat modeling assumptions, detection logic, and reporting completeness. The results show convergence around a limited set of recurring attack models and reusable defense patterns. Both machine learning and deep learning, along with detection-oriented, trust-based, and optimization techniques, appear across multiple security themes rather than being confined to a single problem framing. However, threat specification, detection target definition, and trust design reporting are often incomplete or inconsistent, which limits reproducibility and cross-study comparison. Evaluation orientation, including performance metrics and experimental validation practices, is identified as an important direction for future work.

AI-driven digital transformation: a framework for organizational capability assessment and strategic decision-making in technology management

2025-11-26T06:24:08+00:00

This study develops an Agentic AI-driven framework to address critical challenges in digital transformation, including subjectivity. A Dynamic Weight Adjustment algorithm, which is based on Deep Reinforcement Learning (DWA-RL), enables adaptive updating of the weights assigned to each evaluation indicator across four capability dimensions: Technology, Organizational, Strategic, and Ecosystem. The empirical validation involved over 8,000 enterprise samples from the World Bank Enterprise Surveys and case studies by MIT. For the training datasets, supplementary synthetic data has been generated by Monte Carlo simulation and Generative Adversarial Networks. The framework achieves 87.3% prediction accuracy, which is 15.8% higher than MIT CISR and 17.5% higher than McKinsey, shows the best dynamic adaptability of 4.6/5.0, and improves the quality of decisions by 28% compared to the traditional experience-based approach. Under volatile environments, the DWA-RL algorithm keeps the decline within 17.6 percentage points, while for fixed-weight methods, the decline is as high as 25.5 points. Manufacturing enterprise transformation trajectories prove balanced four-dimensional capability development over three-year periods. The current study extends dynamic capability theory by introducing mechanisms of autonomous agents and redefining the agent-dominated human-supervised decision paradigm.

Research on AI-enabled collaborative governance mechanism for content security: an optimization perspective of review technology based on deep learning

2025-12-04T10:37:33+00:00

In order to bridge the gap between technological optimization and institutional design in Internet content security governance, an integrated framework was constructed, incorporating deep learning-based review technology and multi-stakeholder collaboration. A methodology leading to a three-layer dynamic coupling governance model covering technology, process, and institution with an extended Stackelberg game framework was developed for formal modeling of the strategic interactions among regulators, platforms, Artificial Intelligence (AI) systems, and users. In this connection, an adaptive cross-modal confidence propagation algorithm was presented to improve the accuracy in reviewing multimodal content, together with a Thompson sampling-based dynamic threshold optimization mechanism. On comprehensive test sets, the accuracy of the dynamic collaboration mechanism was 94.6%, and game equilibrium attainment was 95.8%. Compared with pure manual review, costs were reduced by 76%, and efficiency was increased by 8.7 times. Meanwhile, the cross-modal confidence propagation algorithm showed an accuracy increase of 8.4% in high-uncertainty situations. Cross-scenario generalization capabilities have also been tested and verified on social media, short video, online education, and e-commerce platforms. The proposed collaborative governance mechanism can effectively balance accuracy, efficiency, and cost in content moderation and provide a theoretical basis for AI-enabled governance research.

Edge AI-enabled real-time process control in smart plywood production: IoT integration and intelligent automation framework

2025-12-10T07:48:48+00:00

In response to the technical requirements for real-time quality control in the hot pressing process of intelligent plywood production, this study proposes a real-time process control framework driven by edge AI. This framework employs a three-layer edge intelligence architecture. This work shows a practical and efficient boundary node model application scheme for defect detection with multi-level lightweight strategies. In particular, this work builds a decision level data fusion approach for visual detection data and process parameters based on rules for defect-process parameter association mapping. Experimental results have shown that this designed scheme can efficiently detect defects in an edge computing environment. Additionally, with more multi-source fusion being considered in the site environment, the overall detection efficiency might be improved while maintaining a stable closed-loop control system. After that, quality enhancement for products and efficiency improvement for detection were realized. The results provide a feasible method for utilizing engineering processes for enhanced online quality detection for the plywood hot-pressing process based on practical experiences for intelligence upgrades in wood processing.

Artificial intelligence-supported spatial optimization of urban greenway networks: a framework for enhancing restorative environmental performance

2025-12-15T03:39:28+00:00

The goal of the study was to construct an AI-driven end-to-end framework to improve the restorative environmental performance of urban greenway networks. Generally, methods for greenway planning may have subjectivity, low optimization efficiency, and difficulty in quantifying multi-dimensional objectives. The framework integrates convolutional neural networks (ResNet-50) convolutional neural networks (CNN) for landscape quality assessment, GraphSAGE-based graph neural networks (GNN) for spatial topology modeling, and proximal policy optimization (PPO)-based deep reinforcement learning (DRL) for multi-objective optimization. A system for restorative assessment was established based on Attention Restoration Theory. This system has four dimensions, being away, fascination, compatibility, and extent. The generalization of the framework was systematically validated in the case of three representative urban scenarios; plains of a medium-sized city, hills of a small city, and a high-density metropolis. The findings show that compared to manual planning, the framework yielded a restorative score improvement of 42.2%, a 72.7% increase in population coverage, and 98.1% enhancement in efficiency (Optimization time from 120 hours to 2.3 hours). Spatial equity improved due to the decrease in the Gini coefficient from 0.42 to 0.28. Strong transferability is evident as migration cost in cross-scenarios is under 5%. The performance dropped by 26% to 41% when any of the modules (CNN, GNN, DRL) were removed. Multi-objective optimization was better than single-objective techniques. The framework endowed with quantitative decision-support tools, facilitates healthy city construction. It promotes spatial justice by directing physical resources to the most vulnerable sections of the community. Further, the framework provides support for rapid iterative planning of green infrastructure for a smart city.

A hybrid deep-handcrafted feature fusion framework for image based android malware detection

2025-12-15T07:43:20+00:00

Dynamic loading and code-manipulation techniques that weaken the reliability of traditional static and signature-based detectors. Image-based malware analysis has recently emerged as an effective alternative, as transforming executable bytecode into grayscale images reveals structural, spatial and statistical patterns that remain difficult to conceal. Motivated by this, the present study proposes a hybrid learning framework for Android malware detection using grayscale images generated exclusively from DEX bytecode segments. Multiple deep feature extractors based on Transfer Learning architectures—including DenseNet121, MobileNetV2 and InceptionV3—are employed to obtain high-level semantic representations from DEX images, while handcrafted descriptors such as HOG, SIFT, ORB, LBP and GLCM capture complementary gradient and texture characteristics. The fused feature representations are evaluated using several machine learning classifiers, including Random Forest, Logistic Regression, SVM, KNN and Naïve Bayes. Experimental results demonstrate that the DEX image representation yields highly discriminative patterns, achieving a maximum accuracy of 94.40% with Random Forest and 94.33% with Logistic Regression. These findings confirm the effectiveness of DEX-driven image analysis and hybrid feature fusion as a robust, scalable solution for Android malware detection.

MANET studies as experimental artifacts: a PRISMA-guided review of contemporary evaluation orientation, reporting completeness and reproducibility

2025-08-23T08:03:27+00:00

Experimental evaluation is central to MANET research, yet performance claims are often derived from studies that use heterogeneous and inconsistently reported evaluation setups, limiting cross-study comparability, reproducibility, and interpretability. This paper presents a PRISMA-guided systematic review of MANET evaluation practice published between 1 January 2020 and 12 August 2025, using an evidence-mapping and meta-research synthesis approach. Twenty studies were analyzed using a structured extraction template capturing evaluation orientation, experimental platform, mobility and scenario configuration, baseline selection, metric portfolios, and energy modeling practices. Methodological rigor was assessed using explicit indicators for validation reporting, statistical analysis reporting, and reproducibility support, with a derived rigor score summarizing reporting strength across studies and over time. The results provide study-attributed evidence maps and diagnostic summaries that quantify dominant evaluation orientations, heterogeneity in evaluation stacks, and uneven disclosure of reproducibility-critical details. The paper derives a practitioner-oriented checklist that specifies minimum reporting and evaluation design elements needed to support transparent and comparable MANET experiments. Future research should develop and validate community-aligned reporting and benchmarking standards that reduce evaluation-stack ambiguity and strengthen cross-study synthesis in MANET research.

A multi-model database framework for interoperable IoT sensor data management in smart manufacturing systems

2025-11-29T13:02:09+00:00

The explosive adoption of IoT enabled smart manufacturing has increased the complexity of managing heterogeneous sensor data coming from diverse machines, communication protocols, and vendor-specific formats significantly. Conventional relational and time-series databases are very hard to adapt to the twin problems of high volume of data, structural diversity, and semantic inconsistency in industrial environments today. This paper proposes a multi-model database system in order to achieve high-performance interoperability for heterogeneous IoT sensor streams in the Smart Manufacturing Systems. The architecture includes a semantic integration layer to transform the data coming from formats such as JSON, XML, CSV, OPC-UA, and MQTT into a common canonical data model. The framework is evaluated on a synthetic but realistic Industry 4.0 dataset with roughly 5000 devices and over 40000 sensor measurements that allows the ingestion performance, cross-sensor query latency, and scalability of the framework to be evaluated. Experimental results show increased interoperability, support of unified cross-modal queries and low latency performance under growing loads of data. Furthermore, the cross-sensor correlation and analysis of any anomaly points to the applicability of the framework to analytics-oriented tasks, such as the early detection of abnormal machine behaviour. In general, the offered solution offers a semantically consistent, scalable base of interoperable IoT data management of smart manufacturing settings.

Boosting organic pollutants degradation by surface defects of iron oxide nanofibers

2025-12-26T03:27:07+00:00

The release of organic dyes is a serious concern to the water resources and the quality of the environment, and there is a strong necessity to create effective catalysts that can degrade organic dyes. Here, the nanofibers of iron (III) oxide were produced using the electrospinning technique and the high-temperature firing technique, and the oxygen vacancy defects on the surfaces of the nanofibers were formed using the reduction technique of NaBH4. The findings show that the surface oxygen vacancies significantly enhance the catalytic activity. Specifically, the Fe2O3−VO nanofibers achieved near-complete degradation of MB within 4 minutes, with a reaction rate constant (k≈0.95 min-1) approximately 68 times higher than that of pristine nanofibers. Such optimized properties not only demonstrate the potential of defect-engineered iron oxide in organic dye treatment but also provide insights into the structural design of highly effective catalysts for sustainable water remediation.

DiProMo model: PLS-SEM validation of the link between digital transformation, business model, and productivity

2025-12-19T19:34:10+00:00

This paper validates the DiProMo model, which seeks to establish relationships among Digital Transformation (DT), the Business Model (BM), and Productivity (P), using a sequential mixed-methods approach that integrates a systematic literature review, expert interviews, and a survey of 318 professionals, whose data were analysed using PLS-SEM in SmartPLS 4. The results show that DT has a significant effect on innovation, which is part of the BM (β = 0.783), which in turn generates a positive effect on P (β = 0.755). Likewise, a direct impact of DT on P is observed (β = 0.734), and a robust partial mediation of DT with P through the BM is also evident. The overall model fit (SRMR = 0.069) and high predictive values (Q² > 0.50) support the validity and utility of the proposed model. The results indicate that P increases when digital investment is aligned with strategic design in BM, providing empirical evidence and practical guidance for organisational management and the formulation of DT policies geared towards competitiveness.

Federated learning for breaking data silos in smart governance: a privacy-preserving framework for cross-agency collaboration

2025-12-31T08:26:11+00:00

The realization of smart governance highly relies on the effective integration and collaborative utilization of cross-departmental government data, yet data silos that have formed over time and the privacy compliance risks faced by traditional centralized sharing models severely constrain the improvement of collaboration effectiveness. Addressing this challenge, this study proposes the FedGov privacy-preserving federated learning framework for smart governance scenarios, designing a three-layer system architecture comprising data, computation, and coordination layers to support multi-departmental heterogeneous data collaboration, and developing the FedGov-DP algorithm integrating dual mechanisms of differential privacy and secure aggregation to realize the "data usable but invisible" cross-departmental collaboration paradigm. Systematic experiments simulating government scenarios based on public datasets demonstrate that the proposed framework effectively breaks down data silos and achieves significant collaboration gains, the differential privacy mechanism effectively defends against membership inference attacks, and the method exhibits good adaptability to moderate data heterogeneity common in government scenarios. This study extends the application boundaries of federated learning in the public governance domain, provides a new technical pathway for addressing the government data silo dilemma, and the constructed framework with parameter configuration guidelines offers technical support and practical reference for smart governance digital transformation.

Impact of transmission power on safety message communication under sparse vehicular ad hoc networks

2025-12-04T08:28:23+00:00

Transmission power is an important determinant of the performance of Vehicular Ad Hoc Networks (VANETs) due to its direct influence on the reliability and efficiency of safety message communication in both vehicle-to-vehicle and vehicle-to-roadside unit encounters. Vehicular mobility-induced dynamic topology makes it difficult to maintain stable connectivity, especially in sparse and intermittently connected network environments. As a result, resorting to fixed transmission power levels leads to network performance and connectivity degradation. This paper presents a novel model for connectivity assessment that aims to study the effect of varying transmission power levels under various VANET scenarios. The model evaluates network performance across multiple transmission power configurations and traffic densities using key efficacy metrics, including connectivity stability, communication overhead, latency, and safety message delivery performance. The results demonstrate that inappropriate transmission power selection negatively affects VANET connectivity by increasing channel contention in dense regions and significantly reducing communication reliability in sparse and void areas.

Emotionally intelligent social robot for dementia care: empathy-based conversational intervention model using multimodal sentiment analysis

2026-01-08T18:24:38+00:00

Dementia is a challenging health issue for healthcare systems across the globe. Communication disabilities and behavioral changes have been significantly impacting the well-being of patients with dementia. The study formulates an empathy-based conversational intervention approach for patients with dementia using multimodal sentiment analysis. The proposed system applies a cross-modal attention-based model to synthesize speech, facial expressions, and biological signals for effective emotion identification. The synthesis is further augmented with an innovative large language model-based conversational response generation module that can develop appropriate empathetic responses. Experiments conducted on public benchmark databases confirm that the trimodal fusion-based model outperforms state-of-the-art methods with an overall weighted average accuracy of approximately 87.3% for emotion identification. The proposed approach outperforms state-of-the-art methods such as MulT, MISA, and MAG-BERT. The scores on human evaluation of the generated empathetic dialogue reached 4.12 and 4.28 in terms of empathy and coherence, with improvements of 17.0% and 12.3% over baseline models. The meta-analytic synthesis of previous clinical evidence revealed significant beneficial effects of social robot interventions on depression, loneliness, and agitation of people with dementia. The comparison with commercial models such as PARO, Pepper, and NAO showed the superiority of the proposed approach over others in terms of multimodal emotion recognition and dialogue adaptability. These results show that socially interactive robots with high emotional intelligence, equipped with cutting-edge affective computing and natural language processing, have great potential for enhancing the quality of dementia care through personalized emotional support.

Development and evaluation of chitosan–silk fibroin hydrogels for controlled drug delivery in wound healing applications

2026-01-18T12:50:37+00:00

The future of biopolymer-based hydrogel formation holds a promising future for wound healing applications in terms of a controlled drug delivery system. This paper combined chitosan (CS) and silk fibroin (SF) in order to produce a hydrogel with an improved antibacterial effect, mechanical stability, and biocompatibility. A natural cross-linker, genipin, was used, providing tunable mechanical strength with cytocompatibility. Tensile testing, swelling, structural, and thermal tests of the hydrogels were conducted using the Fourier transform infrared spectroscopy (FTIR), differential scanning calorimetry (DSC), and X-ray diffraction (XRD). Structural and thermal properties of the CS-SF hydrogels were investigated. Model antibiotic and anti-inflammatory agents were encapsulated, and in vitro release was measured, demonstrating a controlled, sustained release profile. Superior cytocompatibility was confirmed through cell viability tests with fibroblasts and keratinocytes, indicating that the hydrogels can be applied to wounds. Further diffusion of the drug was also modelled using COMSOL Multiphysics, and the simulation outcomes were compared with experimental release data and found to be highly correlated. The findings demonstrate that under cross-linked genipin CS-SF hydrogels can successfully serve as wound dressings that could potentially enable tissue regeneration, and controlled delivery of therapeutic agents, with immense clinical translation potential in advanced wound care.

The synergetic effect of in-situ TiB2 and ex-situ Al2O3 in Al-Si binary alloys for the enhancement of wear properties

2026-01-17T12:40:00+00:00

In metal matrix composites (MMCs), non-metallic or inter-metallic phases are commonly introduced into a metal or an alloy in a required proportion with an aim to develop a new material that possesses attractive engineering properties. Aluminium matrix composites (AMCs) have been known for their high strength-to-weight ratio, and these are utilized to meet specific requirements in the field of various automotive and aerospace applications. In this study, the hypoeutectic and hypereutectic Al-Si composites have been fabricated through two step stir casting process. TiB₂ incorporation has been facilitated by an exothermic reaction between the molten metal and halide salts (K₂TiF₆ and KBF₄), leading to the in-situ formation of the reinforcement. To develop hybrid composites, Al₂O₃ particles were further added to the melt. The resulting microstructures were analysed using scanning electron microscopy. The presence of Al₂O₃ and TiB₂ phases was confirmed by X-ray Diffraction (XRD). The composites were also evaluated for hardness, density, and dry sliding wear behaviour. The results indicate that the wear rate decreases with increasing TiB₂ content, whereas this improvement is restricted to the addition of 1 wt. % Al₂O₃. Higher applied loads resulted in increased wear rates. Density values of the composites have been found to increase marginally with higher reinforcement content. Hardness improves with Al₂O₃ addition up to 1 wt. %; and a further addition of 2 wt. % causes a decrease in hardness due to particle agglomeration.

AI-driven assessment of urban greenway restorative environments: integrating deep learning, street view imagery, and environmental psychology

2025-12-25T06:17:41+00:00

Existing methods for evaluating urban greenway restorative environments lack objectivity, efficiency, and theoretical integration. The purpose of this research is to develop a restorative environmental assessment framework for urban road binding using deep learning, street-view image data, and environmental psychology theory. It uses a semantic segmentation model called DeepLabV3+ to collect six visual environment features which are otherwise difficult to represent numerically. At the same time, it offers a methodological path for the interdisciplinary integration of artificial intelligence technology and environmental psychology theory. The calculation model of the comprehensive recovery index is constructed in four dimensions based on attention recovery theory. According to empirical analysis, this framework can successfully identify systematic differences in the restorative dimension of different types of binding paths. The presence of greenness can make a large positive contribution to the restorative effect, while building occlusions can have an inhibitory effect. The evaluation results are quite consistent with theoretical predictions and have good robustness in parameter Settings. The findings of the study offer a scientific evaluation tool for accurate diagnosis and optimization improvement of the urban road binding restorative environment. At the same time, it offers a methodological path for the interdisciplinary integration of artificial intelligence technology and environmental psychology theory.

How green intellectual capital shapes competitive advantage through AI-driven green innovation: an empirical study in the manufacturing industry

2026-01-20T10:47:58+00:00

Amid increasingly stringent environmental regulations and rapid global digitalization, the role of green intellectual capital in forging competitive advantage for firms in the manufacturing industry has become a growing concern. This study aims to propose an integrated framework for examining the mediating effect of green innovation and the moderating impact of the application of artificial intelligence on the relationship between green intellectual capital and competitive advantage. Grounded in the natural resource-based theory and the dynamic capabilities theory, the research uses data collected from 163 Chinese manufacturing enterprises and applies partial least squares structural equation modeling, with machine learning techniques such as XGBoost-SHAP. The results reveal that all three dimensions of green intellectual capital, namely green human capital, green structural capital, and green relational capital, are positively associated with both green product innovation and green process innovation, with green structural capital exhibiting the strongest effect. Green innovation partially mediates the relationship between green intellectual capital and competitive advantage. The application of artificial intelligence positively moderates the relationship between green intellectual capital and green innovation, particularly through the green relational capital pathway. Notably, machine learning analysis uncovers a threshold effect whereby an artificial intelligence application must reach a critical level before generating positive impacts on competitive advantage. This study contributes to the literature by extending the natural-resource-based view to digital contexts and demonstrating complementary insights through the integration of explanatory and predictive analytical approaches.

Machine learning-driven pay gap analysis: predicting corporate innovation performance using XGBoost and SHAP interpretability

2026-01-29T17:03:36+00:00

The relationship between pay gap and corporate innovation has been the focus of significant theoretical discussion, with tournament theory and social comparison theory generating contrasting predictions. Traditional linear methods are ill-suited to capturing the nonlinear nature of the relationship. This study proposes an XGBoost-SHAP approach to predict innovation performance using a sample of 26,815 firm-year observations from Chinese A-share-listed firms from 2010 to 2023. The results show that the XGBoost model achieves an R2 of 0.382, which is 65.4% higher than OLS (R2=0.231). The SHAP value analysis indicates that the vertical pay gap ranks as the third most important factor, following firm size and firm age. The SHAP dependence plot shows an inverted U-shaped relationship between the vertical pay gap and innovation performance, with a turning point at approximately 8.7 times. The heterogeneity analysis indicates that state-owned enterprises attain their turning point earlier (7.2 times) than non-state-owned enterprises (10.1 times), suggesting that employees are more responsive to pay inequality. These findings provide practical insights that may guide managers in designing their firms’ compensation schemes. Firms that fall below the threshold may consider expanding their pay gaps, while those that fall above may consider compressing their pay gaps. This XGBoost-SHAP approach translates statistical evidence into practical diagnostic tools that managers may use to assess the optimality of their firms’ compensation schemes in supporting innovation.

Deep learning-driven regime switching models for capturing structural breaks and volatility clustering in financial time series

2026-01-21T06:07:16+00:00

The mechanism of switching and the identification of structural breakpoints in financial markets across different macro environments have long been core challenges in asset pricing and risk management. Traditional parametric models suffer from insufficient flexibility to capture nonlinear dynamic processes. This study proposes a regime-switching model driven by deep learning. By integrating a bidirectional long short-term memory network with the probabilistic inference framework of the Markov transformation process, a unified optimization framework for structural break identification, market regime classification, and volatility clustering modeling is constructed, in which the attention mechanism dynamically focuses on key historical information to enable mechanism identification. A multi-layer perceptron generates state-dependent GARCH parameters to adaptively capture the characteristics of heterogeneous fluctuations, and adaptive threshold monitoring based on KL divergence enables quantitative identification of structural breaks. Experiments show that the model achieves significant performance advantages over traditional methods in structural break detection, mechanism transition identification, and volatility prediction. The cross-market generalization ability and robustness analysis verify the model's applicability across different asset classes and time horizons. The posterior probability distribution of the model's state output can support asset allocation decisions, and the breakpoint identification mechanism provides quantitative early-warning indicators for regulators. It has important practical value in scenarios such as portfolio management, market timing strategies, and systemic risk prevention.

Ethics of AI-based supply chain optimization: a better balance between efficiency and fairness

2026-01-26T20:11:31+00:00

This study examines the ethical issues related to AI-driven supply chain optimization, such as algorithmic biases, the effects of automation on employment, and accountability and transparency. Given the goal of increasing efficiency, machine learning, predictive analytics, natural language processing, and artificial intelligence (AI) are being actively used in a variety of industries, including retail, healthcare, and logistics. Also, technologies are automating and improving tasks such as inventory tracking and demand forecasting. This lowers cost and increases supply chain flexibility. However, using them raises significant ethical problems, specifically the issue of making fair choices. With the presence of bias in the trained systems, there will be unfair distribution of resources and the conditions that define the consequences of decisions, such as the introduction of high-value goods over fundamental needs, and in this case, the population needs it most. Another important issue is job loss, especially in low-skilled jobs, as automation becomes the norm in the logistics industry. The study suggests that AI systems should adopt ethical principles, such as fairness, transparency, and accountability. It suggests practical steps that businesses should take to employ AI in ways that ensure everyone gets fair results. The study continues by emphasizing the need to be aware of ethical issues to use AI to improve efficiency while also promoting fairness and sustainability in global supply chain management.

Secure migration patterns from Java 8 to Java 17 in the mission-critical ecosystem: a risk-driven approach to modernization

2026-01-27T12:10:21+00:00

Upgrading from Java 8 to Java 17 in the enterprise setting is both challenging and an opportunity, especially for mission-critical ecosystems. Java 17 is a long-term support (LTS) version that includes numerous enhancements over Java 8, including language syntax enhancements, improved garbage collection and memory management, enhanced security models, and modularization via the Java Platform Module System (JPMS). Nevertheless, the opportunities presented by Java 17 can be fully realised with the assistance of a well-organised migration plan that assesses risks, implements mitigation strategies, and ensures the equipping of enterprise-scale systems. This paper proposes a risk-based migration framework that is SAP-based, Java environment-specific, and identifies safe migration patterns, along with a detailed case study example to demonstrate how the workflow migration methodology works. We suggest an expedient and replicable solution that accommodates modernization.

Influence of public transportation on urban mobility in Celaya: a GIS case study

2026-01-17T06:04:32+00:00

Sustainable urban mobility represents one of the greatest challenges for medium-sized Latin American cities, where public transport plays a fundamental role in reducing air pollution and traffic congestion and improving access to basic services. This paper examines the impact of public transportation on urban mobility in Celaya, Guanajuato, using Geographic Information Systems (GIS) to study the public transportation network in terms of coverage, accessibility, and efficiency. By compiling georeferenced data on public transport routes, such as departure frequencies and passenger flows, together with details on road infrastructure and sociodemographic data from open sources. Geographic information systems were used to construct thematic maps and spatial accessibility models, which made it possible to identify areas with poor coverage, long travel times, and disparities in urban connectivity. The findings show that, although public transport covers most of the urban areas, there are peripheral areas with poor accessibility and high dependence on private transport, which negatively affects sustainable mobility in developing industrial cities. Likewise, strategic corridors were identified where improving frequencies and modal integration would significantly increase the efficiency of the system. Finally, it is essential to include spatial analysis through GIS in the design of public transport, as this enables fairer and more sustainable mobility policies to be implemented, helping to reduce congestion and improve the quality of life in cities.

Multi-agent reinforcement learning for global virtual power plant collaborative scheduling: a new approach to optimizing renewable energy consumption

2026-02-10T08:07:51+00:00

The integration of high-penetration renewable energy sources (RES) into global power systems necessitates advanced scheduling strategies to ensure supply-demand balance. Virtual Power Plants (VPPs) serve as critical aggregators for distributed resources; however, coordinating VPPs across multiple regions is hindered by the curse of dimensionality, partial observability, and stochastic volatility. Conventional centralized optimization lacks scalability for real-time applications, while single-agent approaches fail to effectively address complex collaborative dynamics. To overcome these limitations, this paper proposes a collaborative scheduling framework based on Multi-Agent Reinforcement Learning (MARL). We model the global system as a multi-regional environment where heterogeneous agents operate under a Centralized Training with Decentralized Execution (CTDE) architecture. A composite reward function is designed to balance economic efficiency with RES absorption, utilizing an attention-based mechanism to exploit time-zone complementarity. Simulation results demonstrate that the proposed method significantly outperforms baseline strategies. Specifically, it achieves a global RES accommodation rate of 94.2% and maintains a minimal tie-line violation rate of 0.8%, compared to only 76.5% accommodation with rule-based heuristics. Furthermore, the approach exhibits superior robustness in extreme-volatility scenarios where standard methods degrade. This study validates the efficacy of distributed intelligence in solving large-scale energy dispatch problems, offering a scalable and privacy-preserving pathway for managing the Global Energy Interconnection.

An affordable mobile LiDAR approach for efficient three-dimensional analysis of urban traffic

2026-01-15T07:08:02+00:00

Urban mobility management is an issue that smart cities cannot ignore, and it requires reliable, sustained, and precise dynamic monitoring of traffic flows. This paper introduces a cost-effective mobile LiDAR-based methodology for three-dimensional urban traffic analysis, providing the high-resolution spatial data necessary for future AI-driven mobility decoding. We use real-world data acquisition rather than conventional studies that rely on traffic simulation tools, such as VISSIM or AIMSUN, to model traffic dynamics, including vehicle volumes, vehicle shapes, inter-vehicle distances, and automatic vehicle counting. The LiDAR system was a mobile system that used a terrestrial laser scanner (TLS) to capture high-density 3D point clouds at various urban intersections with no heavy infrastructure. The suggested methodology encompasses the whole processing chain, i.e., data collection, preprocessing, object segmentation, vehicle localization, volume estimation, and infrastructure element localization. The experiment at two intersections in the city of Tangier (Morocco) demonstrates that the obtained real-world LiDAR data is comprehensive, visually accurate, and suitable for training artificial intelligence models for traffic analysis and management. The proposed workflow provides a foundation of geometric data that could be used for future AI-based traffic analysis, following further annotation and model development.