Future Technology

Analyzing process variables for WEDM of Nimonic alloy 75 with a cryogenic treated tool

Saidulu G, P Prasanna — Wed, 24 Sep 2025 00:00:00 +0000

The present experimentation employs a Wire Electric Discharge Machining (WEDM) technique to investigate how various operational limiting factors influence Material Removal Rate (MRR), Micro Hardness (MH), and Vertex Angles (VA). Nimonic Alloy 75 sheets were used as the raw material for the experiments. Two types of tools were utilized: cryogenically treated brass wires and non-cryogenically treated brass wires. The primary process parameters analyzed in this research include the tool electrode, Ton, Wire Feed rate (WF), Wire Tension (WT), and Toff. The wire diameter was kept uniform at 0.25mm, as was the thickness of the work material Nimonic Alloy 75. The study compares MRR, MH, and VA when using a cryogenically treated tool versus a non-cryogenic tool, considering Ton, WF, WT, and Toff. The experimentations were structured with the help of a Taguchi L-9 OA, and an ANOVA was used to determine the maximum contribution of the variables: vertex angle, microhardness, and MRR. The microstructure of the machined samples, using untreated and CT brass wires, was examined with a Scanning Electron Microscope (SEM). Furthermore, chemical analysis was performed using EDS, comparing weight percentages before and after treatment.

A dual-resource rooftop system for water-energy sustainability: a case study at Near East University’s grand library

Fri, 26 Sep 2025 00:00:00 +0000

The present paper introduces a novel rooftop hybrid system that integrates a solar power system (SPS) and a rainwater harvesting system (RHS) at the Grand Library of Near East University in Northern Cyprus. In this study, ground-based rainfall measurements from 2015 to 2023 were compared with six satellite-derived precipitation datasets (CHIRPS, CFSR, ERA5, ERA5-AG, ERA5-LAND, and MERRA2) to determine the potential for rainwater collection. The results demonstrate that the ERA5-AG dataset produced the highest accuracy based on the values of key metrics (R-squared, Root Mean Squared Error, and Mean Absolute Error), and was used to estimate rainfall harvesting potential. Based on maximum and average daily rainfall, the winter season provided the most potential, with up to 1,665.12 m³ and 373.51 m³ of collected rainwater, respectively. Moreover, the environmental and economic viability of the proposed RHS-SPS system is evaluated through mathematical modelling. The results indicate that a tilt angle of 34° and a north-oriented face generated the highest annual energy output (735,316.3 kWh) and the best capacity factor (19.52%) among the various orientation angles. Furthermore, the Levelized Cost of Energy (LCOE) for the system varied from 2.860 to 5.503 cents/kWh. Besides, a 34° tilt produced the most reduction in CO2, according to analysis of the environmental assessment. According to the findings, the RHS-SPS system aims to address two critical challenges in semi-arid regions (renewable energy production and water conservation) and achieve the Sustainable Development Goals.

Multi-source field sensor data fusion based on cross modal attention mechanism and reinforcement learning driven pesticide application optimization model: towards sustainable crop protection

Minkuan Zhang — Mon, 06 Oct 2025 00:00:00 +0000

The intensification of global agriculture demands precise and sustainable pest management strategies, as indiscriminate pesticide application continues to cause environmental degradation and reduce crop resilience. Existing approaches often rely on unimodal sensing or static rule-based spraying, which fail to capture the heterogeneous and dynamic nature of crop-pest-environment interactions. To address this limitation, we propose a multi-source field sensor data fusion framework that combines a cross-modal attention mechanism with a reinforcement learning-driven model for optimizing pesticide applications. The method integrates Unmanned Aerial Vehicle (UAV) hyperspectral imagery, soil and weather sensors, and pest monitoring signals through adaptive attention, encodes temporal dynamics with recurrent structures, and optimizes spraying actions via a PPO-based policy network. Experiments across rice, maize, and soybean datasets demonstrate superior performance, achieving the lowest RMSE (0.162), highest spray precision (88.3%), and notable pesticide reduction (18.3%) compared with state-of-the-art baselines. These findings highlight the potential of cross-modal AI and adaptive control to advance sustainable crop protection, providing a scalable paradigm for intelligent agriculture.

Innovative approaches to software defect prediction using ensemble learning models

Wed, 15 Oct 2025 00:00:00 +0000

Software defect prediction (SDP) is one of the most critical aspects of software quality improvement and efficient use of testing resources. Traditional machine learning models tend to lack both generalizability and performance, especially when faced with imbalanced or small datasets. To overcome these limitations, the current research proposed a stacked ensemble learning model that combines Random Forest, Gradient Boosting, and AdaBoost as base learners, and logistic regression as a meta-learner. A selected collection of 500 software modules was sampled out of four benchmark repositories: CM1, PC1, JM1, and KC1. Stratified sampling, Min-Max normalization, SMOTE-based class balancing, feature selection via Recursive Feature Elimination (RFE), and mutual information ranking were used as preprocessing steps. The training of the models used 10-fold cross-validation, and hyperparameter optimization was done using Grid Search. The findings showed that the stacked ensemble performed better than any single classifier on all measures, with the highest accuracy of 0.88 and statistically significant improvements in precision, recall, and F1-score (p < 0.05). Data balancing and feature selection methods also increased model stability and interpretability. In summary, the suggested framework will provide a powerful, scalable, and resource-optimal system to predict software defects. This method can be replicated in future studies on larger datasets and with deep learning–based meta-models to improve adaptability. Its integration of Recursive Feature Elimination and mutual-information feature ranking within an optimized stacking design, applied to NASA repositories for the first time, demonstrates measurable improvements in generalization and robustness.

Reconstruction of knowledge worker performance evaluation system in the ChatGPT era: an exploratory study based on human-AI collaborative work model

Zhixin Yu, Zhicheng Yu — Fri, 17 Oct 2025 00:00:00 +0000

The emergence of ChatGPT in November 2022 disrupted practice in knowledge work and defied performance-measurement systems in human-exclusive task accomplishment under unprecedented comparability. This current study fills the gap in the literature between traditional models of appraisal and AI-enabled workspaces through the development of an evidence-based model of measuring performance in human-AI collaborative settings. Drawing on systematic analysis of 5,000 LinkedIn job adverts and 2,000 Indeed salary information between 2022-2024, the present study examined the shift in performance needs and skill needs in knowledge sectors following the release of ChatGPT. The study's findings indicated that AI skills are especially needed in 27.8% of knowledge workers' jobs, with a growth rate of 376% since the release of ChatGPT. AI-trained staff are rewarded with a 17.7% overall premium for their wages, and occupational competence varies from 43.2% in high-tech to 9.7% in the public sector. Systematic skill differences cannot be captured by conventional measuring systems, according to the results. The study discovers a three-dimensional model for measuring performance, including AI Tool Mastery, Collaborative Work Quality, and Human-AI Synergy to measure hybrid skills developed through human-machine collaboration. The research establishes the theory of performance management by developing operational measurement solutions for companies going through workplace redesign due to AI.

Optimized cycle time forecasting in semiconductor wafer fabrication via hierarchical transfer learning and hyperparameter optimization

Kanaparthi Anil Kumar, K. Hemachandran — Sat, 18 Oct 2025 00:00:00 +0000

Accurate cycle-time forecasting remains a persistent challenge in semiconductor wafer fabrication due to highly dynamic, multivariate process conditions. This study proposes an optimized Hierarchical Transfer Learning with Hyperparameter Optimization (HTL-HPO) framework that integrates cross-fab knowledge transfer with Bayesian Tree-Structured Parzen Estimator–based optimization to improve predictive precision and generalization. The methodology involves hierarchical pretraining on source fabs, Maximum-Mean-Discrepancy–driven domain alignment, and probabilistic hyperparameter tuning for fine-grained adaptation to target lines. Using a real industrial multivariate dataset, the model’s performance was benchmarked against established baselines—Decision Tree, GRU, and LSTM—under consistent experimental protocols. The proposed approach achieved the lowest forecasting error (MSE = 0.006; RMSE = 0.079) and the highest explanatory power (R² = 0.934; Explained Variance = 0.938), with paired t-tests (p < 0.05) confirming statistically significant gains. Results reveal that hierarchical knowledge reuse and Bayesian optimization jointly enhance model stability, convergence speed, and robustness under noise and domain shifts. The findings underscore substantial operational implications for predictive scheduling, resource allocation, and sustainable production within smart-fab ecosystems. Overall, HTL-HPO offers a scalable, interpretable, and deployment-ready framework for next-generation intelligent manufacturing.

Multimodal emotion recognition-driven personalized digital therapeutics for anxiety management

Lusha Zhu, Jinho Yim — Sat, 18 Oct 2025 00:00:00 +0000

Anxiety disorders are among the most widespread mental health challenges, yet conventional treatments face barriers of accessibility, cost, and reliance on subjective measures. Digital therapeutics offer scalable solutions, but current systems lack real-time emotion monitoring and adaptive personalization. To address this gap, this study proposes a multimodal emotion recognition-driven framework for personalized anxiety management. The framework fuses electroencephalography, heart rate variability, facial expression, and speech features via cross-modal attention, and employs a reinforcement learning–based decision engine to dynamically select interventions such as breathing exercises, mindfulness, or cognitive reframing. Adaptive feedback further tailors interventions to user responses. Experiments on DEAP and WESAD datasets showed superior performance over unimodal and traditional fusion baselines, with accuracies of 86.2% and 84.7% and AUROCs of 0.91 and 0.89. Anxiety reduction analysis demonstrated up to 24% improvement in State-Trait Anxiety Inventory scores. The study advances affective computing by linking multimodal sensing with adaptive therapeutic design, and offers a foundation for scalable, interpretable, and clinically relevant digital mental health interventions.

Fin orientation effect on passive cooling of photovoltaic panels: an experimental study under extreme hot climate

Fri, 24 Oct 2025 00:00:00 +0000

Solar photovoltaic (PV) panels are a fast-growing solar technology worldwide. However, the PV efficiency is still limited, especially in hot locations due to increased operating temperature. In this research, a PV panel cooled using L-shaped aluminium fins attached passively in various orientations was tested and analyzed compared with another conventional PV under hot Iraqi weather conditions. The average surface temperature reduction, power output augmentation, and electrical efficiency improvement were analyzed and discussed. The research outcomes exhibited positive thermal advancements for the modified PV regardless of fin orientation, with superior performance for the random arrangement. The PV maximum average surface temperature was reduced by 6.5 °C for the random fin arrangement, utilizing the vortices generated from various fin directions. The power output of the modified PV panel was improved over the conventional one by up to 6.54, 10, and 17 W for the vertical, horizontal, and random arrangements, respectively. Besides, the electrical efficiency of the PV with random fin orientation was augmented by 8.6 %, 13.7 % and 23.1 % compared to the vertical, horizontal, and base PVs.

Multimodal fusion and AI context awareness in smart kitchens: deep learning for personalized recommendation and real-time monitoring

Jiaying Li, Jinho Yim — Sat, 25 Oct 2025 00:00:00 +0000

The proliferation of artificial intelligence (AI) and the Internet of Things (IoT) has positioned smart kitchens as a frontier for innovation in personalized nutrition, safety monitoring, and sustainable consumption. Despite rapid progress, existing approaches remain fragmented: vision-based systems struggle with occlusion, speech-driven interfaces are vulnerable to noise, and IoT sensor networks, while reliable, often lack semantic integration with user preferences. Personalized recommender systems further suffer from static designs that fail to adapt to evolving contexts. Addressing these limitations, this study introduces a multimodal deep learning framework that unifies cross-modal attention and reinforcement learning to achieve context-aware personalization. Visual, auditory, and sensor streams are embedded into a shared representation, fused via attention mechanisms, and subsequently optimized through a reinforcement learning agent that balances nutritional goals, user satisfaction, and safety requirements. Empirical evaluation across three multimodal datasets demonstrates significant improvements over strong baselines, with gains of +8.4% in Top-1 accuracy, +14.0% in F1-score for safety monitoring, and a 23.5% reduction in nutritional prediction error. Interpretability modules employing SHAP and Integrated Gradients further provide transparent explanations, enhancing trust and accountability. The findings underscore the practical value of the framework in promoting healthier diets, improving energy efficiency, and ensuring domestic safety, while laying the groundwork for future applications in healthcare, adaptive living, and sustainable human-AI interaction.

GI meets AI: Glycemic index in the age of AI, computational breakthroughs

Sun, 26 Oct 2025 00:00:00 +0000

This review explores the avenues for the application of Artificial Intelligence (AI) techniques in Glycemic Index (GI) related research. The necessity of sophisticated technologies to investigate various GI‐related studies in food analytics has been established in recent years. AI technologies have emerged as promising approaches to address these challenges. We identified six major AI technologies applied in GI research: Machine Learning, Reinforcement Learning, Deep Learning, Image Processing, Natural Language Processing, and Explainable AI. Some of our findings include: (a) There have been significant improvements in GI-related studies using AI technologies over the past decade. (b) Machine learning algorithms were widely used (c) Many researchers used custom datasets, with the predominance of research originating from North American countries. (d) Identification of limitations and future directions for GI‐related studies employing AI technologies. By embracing AI technologies, the field of food analytics is poised for substantial advancements in understanding and managing glycemic responses. Unlike existing reviews that mainly discuss nutritional or clinical aspects of the glycemic index, this study systematically examines the integration of AI and machine learning technologies in GI-related research. It highlights computational breakthroughs, methodological trends, and future directions for intelligent glycemic analysis.

Integrating SIWEC and Koch snowflake fuzzy sets to prioritize trust factors in an artificial intelligence-based audit system

Feyza Dereköy, İpek Yaylalı, Serhat Yüksel, Serkan Eti, Hasan Dinçer — Wed, 29 Oct 2025 00:00:00 +0000

This study aims to identify effective strategies to increase confidence in AI-based audits. A novel decision-making model is being developed to identify these strategies. In this process, seven criteria are identified through a literature review. Furthermore, opinions on these criteria are obtained from 10 different subject-matter experts. The significance ratio for these people is computed based on their work experience. In this process, an artificial intelligence-based approach is taken into consideration. Furthermore, the weights of the selected criteria are determined using the SIWEC methodology. On the other hand, Koch snowflake fuzzy sets are introduced in this study to address uncertainty in decision-making analysis. Perceived change in audit quality (PCAQ) is the most important indicator, with a weight of 0.181. In addition to this issue, stakeholders’ acceptance and resistance to technology (SART) play a crucial role in this process, with a weight of 0.166. This study contributes to the literature by creating a novel model to identify prior strategies to improve trust in the AI-based audit systems. These findings pave the way to take appropriate actions to increase the effectiveness of this process.

Pictogram semantics standardization for barrier-free drug packaging: deep-learning-assisted design guidelines

Hui Li, Verly Veto Vermol, Zulimran Ahmad — Sun, 02 Nov 2025 00:00:00 +0000

As the world gets older, elderly users find it harder to understand information on medicine packaging. This study created a framework to improve visual communication for older people using deep learning to standardize icons. The research involved 200 participants aged 60 and older who answered questionnaires and took part in interviews, while deep learning models were trained with 1,500 medicine icons. The Residual Network-50 (ResNet-50) model reached 94.8% accuracy, outperforming VGG-16 (89.6%) and Vision Transformer (92.1%), in recognizing meanings across 21 icon types. Analysis showed that performance risk, psychological risk, and safety risk affect how older users accept these icons, with distrust playing a role (R²=0.723), and psychological risk being responsible for 54.6% of the indirect effect. Testing showed that using standardized icons raised recognition accuracy from 68.3% to 92.5% and cut down comprehension time by 52%(t=9.87, p<0.001, Cohen's d=2.21). The recommended design standards (icon diameter ≥20mm, font size ≥14pt, contrast ratio ≥7:1) give measurable guidelines for the medicine industry and are important for encouraging healthy aging.

Exploring various neural network configurations for the NN- based MPC in Multiagent System

Piyush Chaubey, Anilkumar Markana, Dhaval Vyas, Deepak Kumar Goyal — Sun, 02 Nov 2025 00:00:00 +0000

Multi-robot cooperation, unmanned aerial vehicle (UAV) formation control, intelligent transport systems, and distributed sensor networks are just a few domains where multi-agent systems are crucial, as they require coordinated behavior to achieve common goals such as exploration, resource allocation, distributed sensing, and target tracking. This paper investigates various neural network configurations utilized in the NN-MPC framework for consensus control of multi-agent robotic systems. The NN-MPC control is applied to the consensus problem of a leader-follower multi-agent system, where agents coordinate to achieve collective behavior. In this approach, MPC is utilized to predict the future values of the control objective, which is optimized by minimizing a cost function with various neural network architectures. Different neural network configurations based on feed-forward, recurrent neural networks, Fitnet, and cascade networks are explored for the NN-MPC-based multi-agent systems. The analysis is performed through a simulation-based model of a quadrotor fleet system. Results show that the follower agents achieve consensus 60% faster than with RNN-MPC in comparison to the feedforward neural network, whereas the results are more effective when compared with the cascade network configuration-based MPC, where agents reach consensus 90% early if paired with suitable training structures. Overall, the article contributes to the recent topic of research on learning-based MPC of the multi-agent system in achieving consensus for the leader-follower strategy.

AI-enabled factors influencing cultural heritage conservation and tourism development towards tourist experience quality

Ying Long, Daranee Pimchangthong, Kang Li — Mon, 17 Nov 2025 00:00:00 +0000

This research examines the impact of AI technology on the quality of tourist experiences at cultural heritage sites, utilizing an integrated Technology-Organization-Environment (TOE) framework. Analyzing 200 UNESCO World Heritage Sites with 52,847 reviews (2020-2024) using Structural Equation Modeling, we found AI creates dual value pathways: conservation technology enhances heritage value (β=0.45, p<0.001), which strongly influences experience quality (β=0.51, p<0.001), while tourism technology strengthens immersive experiences (β=0.58, p<0.001), which also enhance quality (β=0.36, p<0.001). Both paths significantly improve tourist experience quality, with direct effects of β=0.21 (p<0.01) and β=0.34 (p<0.001) respectively. The integrated model explains 59% of experience quality variance (R²=0.59), superior to alternative specifications. Multi-group analysis reveals technology readiness significantly moderates direct effects (Δβ=0.24-0.25), with sophisticated visitors showing 2-3 times stronger responses, while heritage value appreciation remains universal across digital literacy levels. Findings demonstrate AI enhances rather than diminishes authenticity, with cognitive-emotional appreciation surpassing technological immersion in driving satisfaction.

Trust and adaptiveness enhancements to PRFDRA for secure metaheuristic path selection in MANETs

Wed, 19 Nov 2025 00:00:00 +0000

In Mobile Ad Hoc Networks (MANETs), the Power-Aware River Formation Dynamics Routing Algorithm (PRFDRA) enhanced energy efficiency by forming power-aware paths and facilitating multi-flow diffusion. It remained vulnerable to internal misbehavior. RFDTrust added trust metrics to mitigate malicious activity, but applied them only in neighbor selection along downhill gradients. This limited path diversity and adaptiveness. This paper proposes TA-PRFDRA (Trust-Adaptive-Power-Aware River Formation Dynamics Routing Algorithm), a trust- and adaptiveness-enhanced version of PRFDRA. TA-PRFDRA integrates trust evaluation into all routing stages. It applies dynamic switching between Single Flow Direction (SFD) and Multi-Flow Direction (MFD) based on trust-weighted gradient variance. The algorithm utilizes a composite trust model that takes into account node energy reliability, packet forwarding behavior, route participation, and delay consistency. Trust is applied in gradient, erosion, altitude, sediment transport, and path cost computations. Simulation results show that, compared with PRFDRA, RFDTrust, RFDManet, and TORA, TA-PRFDRA achieved up to 1.33% higher packet delivery ratio (PDR). Average end-to-end delay (AE2ED) decreased by 0.14 s. Detection rate (DR) increased by up to 30.38%. Energy consumption (EC) was reduced by up to 15.94 J. Statistical analysis confirmed that improvements over RFDTrust were significant. These results demonstrate that integrating trust into all routing processes with adaptive flow control enhances reliability, latency performance, security, and energy efficiency in MANETs.

AI-enabled toward zero-emission buildings and clean mobility: PV–BIPV and battery storage integration: a case study of Diyala, Iraq

Wed, 19 Nov 2025 00:00:00 +0000

Iraqi buildings continue to rely heavily on fossil fuels, which raises carbon emissions and energy costs. To address this knowledge gap, the primary objective of the present study is to assess the techno-economic and environmental performance of solar energy retrofitting for a two-story mixed-use building in the eastern Iraqi province of Diyala, utilizing ERA5 reanalysis data for the first time. To this aim, three retrofit scenarios are considered ((1) the baseline scenario (BS) with no renewable systems, (2) the second scenario (SS) with a rooftop photovoltaic (PV) system, and (3) the third scenario (TS) combining rooftop PV, building-integrated photovoltaic (BIPV) glazing and a 30 mm layer of Expanded Polystyrene (EPS) insulation). The simulations were conducted with and without battery storage (103.2 kWh capacity) to demonstrate grid independence and energy self-sufficiency. The findings demonstrate that the TS scenario achieved net-zero or carbon-positive operation, as evidenced by the reduction of annual CO₂ emissions from 39,122 kg (BS) to –9,257 kg (TS), which represents net export of renewable energy to the grid. Economically, SPP ranged from 3.2 to 5.4 years without a battery and from 10 to 14 years with one, and LCOE ranged from 0.038 to 0.072 USD/kWh, demonstrating long-term viability. Furthermore, 90–120 electric vehicles might be charged each month using the extra daylight energy, encouraging sustainable mobility. This study shows that it is possible to create zero-emission buildings that use integrated PV and BIPV systems to allow EV charging, improve grid stability, and lower CO₂ emissions all at once. Besides, the innovative potential of integrated PV-BIPV-battery systems for zero-emission buildings to decarbonize Iraq's urban energy infrastructure is demonstrated in this study.

Deep stacked autoencoder with fractional VCROA for DDoS attack detection using a big data approach in the MapReduce framework

Rahul Vijay Kotawadekar, Suhasini Vijaykumar, Priya Chandran — Fri, 21 Nov 2025 00:00:00 +0000

The rising dependence on internet-based services has exposed network infrastructure to increased vulnerability to cyberattacks, especially DDoS attacks. The attacks flood target systems with unwarranted traffic that disrupts legitimate access and undermines service reliability. To overcome this issue, the present paper proposes an optimization-based deep learning model, called Fractional Velocity Contour-based Remora Optimization Algorithm-Deep Stacked Autoencoder (FVCROA_DSA), for high-efficiency DDoS attack detection in a MapReduce environment. The model combines a mean-substitution method for filling data gaps and Support Vector Machine Recursive Feature Elimination (SVM-RFE) in the mapper step to identify the most significant network attributes. This step is followed by the reducer stage, which trains a Deep Stacked AutoEncoder (DSA) to recognize attack patterns, which is then fine-tuned by the proposed FVCROA algorithm. Fractional Calculus leads to increased optimization stability and faster convergence during training. Experimental tests on the BOT-IoT and DDoS Attack datasets show that the FVCROA architecture with DSA achieves higher detection accuracy, with a precision of 93.857, a recall of 94.827, and an F-measure of 94.340, surpassing the current baseline techniques in scalability and reliability.

Adaptive AI systems and organizational resilience: a multi-level analysis of digital mindset, decision autonomy, and strategic performance

Baoqing Yu, Nur Sayidah, Bambang Raditya Purnomo — Sat, 22 Nov 2025 00:00:00 +0000

This paper examines how adaptive AI systems influence organizational resilience during the COVID-19 pandemic, specifically through the mediating role of the digital mindset and decision-making autonomy. Based on dynamic capabilities theory, the paper develops an innovative conceptual framework that recognizes adaptive AI systems within an integrated technological system that supports the organization's sensing and response capacities during a crisis. Using the Flash Eurobarometer 486 survey conducted in April and May 2020, this study collected data from 12,108 SMEs across 27 European Union member states. The direct effect, mediated relationship, and cross-level interaction strategies employed hierarchical linear models and bootstrap-mediated models with 5,000 iterations. The empirical evidence reveals a significant positive relationship between AI systems and organizational resilience, reducing the odds by 2.342 times (p<0.001) and explaining large incremental variance over the classical organizational characteristics. Digital mindset demonstrated a stronger mediating effect (indirect effect β = 0.17, 95% CI [0.12, 0.24]) compared to decision-making autonomy (indirect effect β = 0.11, 95% CI [0.06, 0.18]). The organizational path-levels and moderations provide critical contextual dimensions, reflected in industry digital intensity, γ=0.15, p<0.05, and national digital infrastructure,γ=0.22, p<0.01. Based on dynamic capability theory, this paper contributes by extending the concept of AI systems to an organizational meta-capability, signposting critical leave-taking measures and implications for managers and policymakers in coping with adverse, turbulence-prone conditions during digitalization within the organization.

Research the association between AI-driven organizational support systems and university faculty work engagement: the moderating role of digital literacy

Zhixin Qian, Andi Tamsang Andi Kele, Ang Hong Loong, Pang Yeng Yuan — Mon, 24 Nov 2025 00:00:00 +0000

According to the Job Demands-Resources (JD-R) model and the Technology Acceptance Model (TAM), this cross-sectional survey examined whether organizational support systems enabled by artificial intelligence (AI) were positively correlated with work engagement among university lecturers and examined the moderating role of digital literacy on 387 teachers at certain Chinese universities. With 9-item multidimensional UWES-9 vigor, dedication, and absorption scale of AI support in teaching, research, and administration domains, hierarchical regression with simple slopes, it was found AI organizational support predicted positively work engagement significantly (β=0.425, p<0.001) and explained additional 18.6% variance after controlling for demographics; digital literacy moderated this highly significantly (β=0.168, p<0.01, ΔR²=0.026), and high digital literacy faculties exhibited 2.35 times stronger strength of relations between AI support and engagement than low digital literacy faculties, and moderation being the highest for vigor dimension (β=0.185); bootstrap analysis with resamples 5,000 and sample split validation confirmed stability of such effects. By conceptualizing digital literacy as a central boundary condition, the current study extends JD-R theory to digital environments and describes another human-AI interaction in which AI complements but does not substitute human capacity and presents empirical evidence of universities to implement all-encompassing digital literacy training programs in parallel with AI system installation, although the cross-sectional study limits causal inference, findings are theoretically meaningful and practically informative and present visionary insight for knowing and promoting faculty well-being in the digital age.

Integrating panel data regression and fuzzy decision-making approaches to evaluate the impact of currency-hedged deposits on participating banks

Selman Duran, Serkan Eti, Serhat Yüksel, Hasan Dinçer — Wed, 26 Nov 2025 00:00:00 +0000

Currency-Hedged Deposits (CHD) were introduced in Türkiye to hedge the currency risk. Hence, it is aimed to provide macroeconomic stability in this country. Nevertheless, the impact of this implication on banks' participation is unclear. This study analyzes the impact of the foreign exchange hedge deposit (CHD) mechanism on the financial performance of participation banks in Türkiye. This study integrates fuzzy multi-criteria decision-making analysis with panel data regression. In this framework, data from these banks for 2021-2023 is considered. First, panel regression analysis is conducted for six participating banks. Second, a Euclidean distance-based CIMAS technique is used to find the most critical criteria. For this purpose, Fermatean fuzzy numbers are considered in this modelling process to handle uncertainties more effectively. The main contribution of this research is the hybrid consideration of panel data regression and fuzzy decision-making analysis. Owing to this combination, the impact of this new implication on bank participation can be more effectively identified. Econometric results indicate that CHD has a positive impact on profitability. On the other hand, risk management and compatibility with interest-free financing are the most critical factors.

Intellectual property protection, digital economy development, and corporate green innovation: threshold effects and regulatory mechanisms

Jun Pan, Rusmawati Said, Normaz Wana Ismail — Fri, 28 Nov 2025 00:00:00 +0000

This study examines the complex interactions among intellectual property protection (IPP), digital economy development, and corporate green innovation using panel data from Chinese listed firms (2014-2022). Employing threshold regression models and moderated regression analysis, we identify significant nonlinear relationships and regulatory mechanisms. Results reveal a U-shaped relationship between IPP and the quantity of green innovation, with identifiable threshold effects across multiple protection regimes. However, IPP exhibits predominantly negative effects on innovation quality across protection levels, with varying intensities observed in different regimes (zones 1-4). Digital economy development demonstrates dimension-specific moderating effects, significantly amplifying the promotional effect on innovation quantity (coefficient 0.711**) but showing minimal impact on innovation quality (-0.085, insignificant), functioning as an "efficiency amplifier" rather than a "quality enhancer." Green agency costs exhibit complex regulatory mechanisms that vary across institutional regimes, resulting in compensatory effects in weak protection environments and triggering institutional overload in strong protection contexts. These findings challenge the linear "more protection equals more innovation" assumption and highlight fundamental distinctions in how technological and institutional drivers affect different dimensions of green innovation. The results have crucial implications for policymakers in designing differentiated IPP regimes and targeted digital economy policies optimized for specific development stages.

Machine learning model for predicting symptom improvement rates in hospitalized deep vein thrombosis patients

Nan Zhou, Teck Han Ng, Chai Nien Foo, Lloyd Ling, Yang Mooi Lim — Mon, 01 Dec 2025 00:00:00 +0000

Deep Vein Thrombosis (DVT) demonstrates considerable treatment response heterogeneity, with 40-60% of patients developing complications despite standard anticoagulation therapy. Accurate prediction of individual treatment outcomes remains an unmet clinical need. This study develops and validates a machine learning-based model to predict symptom Improvement Rate (IPR) using retrospective data from 403 hospitalized DVT patients (2018-2023). Six predictive features are identified using Random Forest-based Recursive Feature Elimination (RFE): age, white blood cell count, Activated Partial Thromboplastin Time (APTT), Thrombin Time (TT), surgical intervention status, and baseline symptom severity. The regularized eXtreme Gradient Boosting (XGBoost) algorithm achieves optimal performance with a test coefficient of determination (R²) of 0.60, Root Mean Square Error (RMSE) of 12.36, and five-fold cross-validation R² of 0.58 ± 0.07. SHapley Additive exPlanations (SHAP) analysis reveals that APTT and surgical intervention are the strongest predictors of treatment response. The validated model is deployed as a publicly accessible web-based clinical decision support tool, enabling real-time outcome prediction at the point of care. This research establishes a practical framework bridging predictive analytics and clinical practice, facilitating evidence-based, personalized DVT management strategies.

Research on real-time data display and production management in a digitalized management factory with an artificial intelligence-assisted flexible manufacturing execution system

ChengHsien Tsai, Oyyappan Oyyappa, Dhakir Abbas Ali — Tue, 02 Dec 2025 00:00:00 +0000

Traditional Manufacturing Execution Systems (MES) face critical limitations in addressing Industry 4.0 demands for real-time processing, flexible scheduling, and adaptive decision-making, with less than 1% of manufacturing data effectively utilized. This research develops an Artificial Intelligence (AI)-assisted flexible MES framework integrating real-time data visualization, digital twin technology, and distributed intelligence to enable proactive manufacturing management. The system employs Design Science Research (DSR) methodology and implements a microservices architecture using Apache Kafka for message streaming, Flink for real-time processing, and TensorFlow for AI inference, deployed across five production lines with 2,350 sensors and 45 Programmable Logic Controllers (PLCs). Results demonstrate exceptional performance with system throughput reaching 12,500 messages per second, the design target by 25%, average data collection latency below 10 milliseconds, and 99.9% availability over 72-hour continuous operation. Production efficiency improved significantly with 25% increased output, 65.7% reduction in defect rates (from 35,000 to 12,000 Parts Per Million), and 87.5% decrease in changeover time (from 120 to 15 minutes). Overall Equipment Effectiveness (OEE) increased from 60% to 82%, approaching world-class benchmarks (>85%). This research validates distributed intelligence architectures for achieving simultaneous improvements in manufacturing flexibility and efficiency, challenging traditional theoretical trade-offs while providing a practical implementation roadmap for digital transformation in manufacturing enterprises.

Computer-aided innovation for intelligent product design: a text mining and knowledge management approach

Wed, 03 Dec 2025 00:00:00 +0000

The fast-paced innovation and the growing need for user-centric products hold traditional design approaches against the wall in the Industry 4.0 era. This research establishes a unified Computer-Aided Innovation (CAI) framework based on text mining, ontology-based knowledge management, and TRIZ-based reasoning to support intelligent product design. The framework uses natural language processing to extract user requirements, technical problems, and potential contradictions from unstructured textual content sources such as product reviews, patents, and technical information. These insights are then structured in a TRIZ-compliant knowledge base to enable the rapid, transparent, and traceable generation of concepts. A smart wearable health device was used as the case study to evaluate the system's performance, and the results showed that the ideation efficiency of all concepts was significantly improved, with all concepts produced in less than 20 minutes, and the results were balanced across novelty, feasibility, and usability metrics. Compared with traditional methods such as brainstorming and Quality Function Deployment (QFD), the proposed framework yielded richer insights, greater concept diversity, and more evidence-based recommendations. Despite these advantages, the approach appears sensitive to textual ambiguity, domain-specific terminology, and the long-term scalability of the ontology repository. Future research will focus on the following areas: leveraging multilingual data sources, combining generative AI with digital twin simulations for time-critical design exploration, and expanding the framework to other product domains. Overall, the proposed CAI framework is part of promoting systematic innovation by incorporating AI-assisted reasoning and structured knowledge representation in the early stages of product design.

Personalized skeuomorphic UI generation for industrial interfaces using diffusion models: a user-centric approach in Chinese manufacturing context

Thu, 04 Dec 2025 00:00:00 +0000

To solve the issue of the digital transformation of Chinese manufacturing in terms of the bottleneck between industrial interfaces not being able to adapt to heterogeneous operators and the high cognitive load imposed on them, the authors propose the SkeuoUI-Gen framework based on the adaptation of skeuomorphic design principles and the use of conditional diffusion models to produce personalized industrial interfaces in the context of Chinese manufacturing. In this regard, the experiment used a within-subjects design involving 250 manufacturing industry operators (diverse in age, experience, and industry sectors) to evaluate three interface types: traditional flat interfaces, fixed skeuomorphic interfaces, and personalized adaptation interfaces. The experiment used objective evaluations (FID and PSNR) and subjective evaluations (SUS score and cognitive load), and trained the model on multiple sources: 50,000 interaction logs from operators and 50,000 screenshots of industrial user interfaces. The experiment found that the personalized adaptation interface resulted in a 78.6% SUS score (an increase of 15.4% compared to the traditional baseline), improved efficiency by 24.7%, and reduced serious safety-related errors by 52% and 67%. The network achieved a lower FID (21.5) than GAN-based approaches and required only 2.3 seconds per generation. In addition, the network presented robustness through multi-dimensional validation. This framework expands the cognitive load theory and the technology acceptance model.

An efficient novel deep learning-based predicting model optimized by an improved DAOA algorithm for microgrid energy management

Ali Q. Almousawi, Nabil Jalil Aklo, Zaid Alhadrawi — Fri, 05 Dec 2025 00:00:00 +0000

This paper presents robust energy-demand and renewable power forecasts for the microgrid using deep learning-based forecasting and a metaheuristic-based optimization model. A Long Short-Term Memory (LSTM) is used to model the temporal nonlinear dynamics of the energy datasets. A new Improved Dynamic Arithmetic Optimization Algorithm (IDAOA) is developed to fine-tune LSTM parameters, incorporating inertial weights, a mutation factor, and the triangle mutation operator to balance exploration and exploitation. The model's performance is verified on various datasets, including wind turbines (WT), photovoltaic (PV) systems, load demands, and day-ahead electricity pricing. This work shows that the IDAOA-LSTM model outperforms other strategies. Practically, the Root Mean Squared Error (RMSE) was 0.021 in the forecast of WT power and 0.031 in the case of PV power. The model performs well in predictions, with high coefficient of determination (R²) values (R² ≥ 0.98) throughout all tasks. These findings strengthen the applicability of the proposed method to enhance energy-saving measures while preserving the stable operation of those microgrid (MG) systems.

Artificial intelligence and digital technologies for piano sight-reading skill development: a scoping review

Ruiqing Rui, Muhammad Syawal Amran, Nurfaradilla Mohamad Nasri — Mon, 08 Dec 2025 00:00:00 +0000

Piano sight-reading is a complex cognitive activity that many pupils remain unable to perform despite sustained educational efforts. AI and digital technology have revolutionized numerous educational fields; however, their integration with educational technology for sight-reading piano remains diffuse and concerning to experts due to a lack of coherence across AI-related investigations. This study aims to systematize knowledge on the application of AI and digital technologies in educational technology for sight-reading piano, following the PRISMA-ScR guidelines. A search of four main databases (Web of Science, IEEE Xplore, Scopus, ACM Digital Library) was conducted for papers on AI-related technology for sight-reading piano from 2014 to 2024. This resulted in screening 368 entries to select 33 relevant to the study objective. Five types of technology exist: AI-related intelligent tutoring systems, computer vision and optical music recognition, pattern recognition with deep learning, applications of virtual reality and augmented reality, and mobile and IoT. The study demonstrates a discrepancy between the complexity of AI and accessibility for pupils. AI-powered tutoring systems and deep learning approaches are showing promising results in controlled settings, but evidence on long-term effectiveness remains limited. A fundamental tension exists between analytical sophistication and accessibility: high-performing systems require substantial computational resources, while accessible mobile solutions provide much weaker analytical capabilities. On the other hand, accessibility for pupils remains a top priority, including the use of IoT technology for educational sight-reading piano.

Research on resilience enhancement mechanism of intelligent supply chain in digital transformation context: synergistic effect of IoT empowerment and edge computing

Zhicheng Yu, Zhixin Yu — Mon, 08 Dec 2025 00:00:00 +0000

This study explores the synergistic effect of Internet of Things (IoT) and edge computing on the supply chain resilience through technological interaction channels. Based on dynamic capability theory and resource coordination theory, the study employs external data sources such as the World Bank Enterprise Survey, the China Industrial Enterprise Database, and the China Ministry of Industry and Information Technology to investigate the research question. Specifically, it uses panel data from 892 manufacturing and logistics enterprises spanning 2020-2024, employing hierarchical regression and simple slope analysis as the empirical methods. The empirical results show that the application level of either IoT technology or edge computing can significantly improve supply chain resilience, with remarkable synergistic effects when the two technologies are jointly adopted. Edge computing can further improve the efficiency of IoT applications by enabling higher application-level thresholds. Additionally, the synergistic effect between IoT technology and edge computing exhibits industrial heterogeneity in optimizing resilience-building efficiency: the manufacturing industry demonstrates a stronger synergistic effect than the logistics industry. This study formally validates the theoretical mechanism underlying technology application, encompassing real-time sensing, edge analysis, and rapid response. It thereby addresses a critical gap in the existing literature and theoretical framework concerning the "resilience-warning capability-response speed" model.

A unified machine learning framework for enterprise portfolio forecasting, risk detection, and automated reporting

Ashutosh Agarwal — Tue, 16 Dec 2025 00:00:00 +0000

This paper proposes and evaluates a unified machine-learning framework for enterprise portfolio management that integrates multi-horizon financial forecasting, unsupervised risk detection, and explainable reporting within a single pipeline. Using a synthetic but structurally realistic ERP-style dataset comprising 162,000 project–month records with 24 financial and operational features, the study adopts a quantitative design based on multi-source feature engineering, expanding-window temporal cross-validation, and benchmarking of five forecasting models (Linear Regression, Random Forest, XGBoost, LightGBM, CatBoost) across 1-, 3-, and 6-month horizons. Hyperparameters for the strongest models are tuned with Optuna, and three unsupervised detectors (Isolation Forest, COPOD, LODA) are applied to scaled numeric features, while SHAP is used to generate global and local explanations. Results show that gradient-boosted trees substantially outperform linear baselines, reducing MAE by roughly 25–40% and achieving R² ≈ 0.63 at 1 month, ≈ 0.57 at 3 months, and ≈ 0.43 at 6 months, with open commitments, backlog, change orders, and schedule slippage emerging as dominant drivers of future spend. The anomaly layer flags around 2% of records as high risk, capturing patterns such as vendor rate spikes, zero-commitment overspend, stalled backlog, and abrupt forecast collapses. Rather than introducing novel algorithms, this work contributes a unified, SHAP-enabled architecture that enhances auditability and governance by transforming model outputs into defensible financial narratives and providing a practical blueprint for future work to extend to real ERP data, streaming architectures, and human-in-the-loop risk governance.

The impact of AR-enabled try-on experiences on consumer purchase decisions: the moderating role of AI-powered recommendation agents

Nuo Cheng — Wed, 24 Dec 2025 00:00:00 +0000

This study investigates how AR try-on functionalities affect consumer purchase behaviors in terms of psychological empowerment processes and uses AI recommendation attributes as boundary conditions. Drawing on skill acquisition theory and dual-process theory, this study hypothesizes a dual-pathway model in which AR interactivity influences perceived control, AR immersion influences perceived value, and both perceptual phenomena influence purchase intention positively. Using the techniques of structural equation modeling and multi-group analysis, data were collected from 500 Chinese consumers through three major e-commerce platforms (Tmall, JD.com, Dewu). The results show that perceived control is a mediator between AR interactivity and purchase intention (indirect effect = 0.406, 95% CI [0.324, 0.495]), and perceived value is a mediator of the relationship between AR immersion and purchase intention (indirect effect = 0.474, 95% CI [0.389, 0.566]). AI-AR integration level significantly enhances the interactivity-control pathway (Δχ2 = 12.87, p <.001), while AI feedback timeliness amplifies the immersion-value pathway (Δχ2 = 10.34, p <.001). These findings imply that the combinations of AR and AI technologies have impacts on consumer decision-making and that the characteristics of AI technology act as boundary conditions. This research has theoretical implications for technology-based consumer empowerment and provides some usable advice on how to better integrate AR-AI technology in online shopping.