his study investigates the short-run causal relationship between public debt (PD) and the interactions of money supply, interest rates, and exchange rate—captured through the composite Money Supply–Interest Rate–Exchange Rate (MIE) index—in five Southern African countries: Botswana, Namibia, South Africa, Zambia, and Zimbabwe, using annual panel data from 2010 to 2025. An understanding of the dynamic feedback effects of the interaction of macroeconomic variables on the dynamics of debt is important. The study employs Granger Causality tests and Vector Autoregressive (VAR) modeling techniques to capture both directional relationships and dynamic feedback effects. The results reveal a bidirectional causal relationship between the MIE index and public debt, suggesting that changes in money supply, interest rates, and exchange rates have a significant effect on debt accumulation, while changes in public debt also affect monetary and financial conditions in the short run. Robustness checks using lagged VAR specifications confirm the stability of the findings. The results of the study have vital policy implications for fiscal and monetary policy coordination for manageable debt sustainability, reduced macroeconomic volatility, and improved economic stability.
The accelerated development and integration of artificial intelligence in the education system requires the use of systematic and evidence-based approaches to assess the state of digitalization. For the publication, a study of the mechanisms for national research and evaluation of digitalization in the education system has been conducted. In most cases, the assessment covers various aspects of digitalization, indicating a lack of an integrated approach that includes all key indicators. The current study bridges this methodological gap by using the “Six Pillars of Digitalization” model as a framework to assess the state of digitalization in the education system. A methodology is proposed that consists of four stages: identifying programs and projects under the six pillars, defining key evaluation questions and performance indicators, selecting data collection methods, and applying the PDCA cycle-based evaluation approach. The National Program “Information and Communication Technologies (ICT) in Pre-school and School Education” is used as an empirical case, as its objectives and implemented activities correspond to the conceptual dimensions of the six pillars. The evaluation results provide a structured overview of the extent to which the national program meets the individual digitalization indicators. The study confirms the Six Pillars of Digitalization as a comprehensive tool for assessing the current state and identifying areas for improvement in digital transformation within the education system.
Customer churn represents a critical challenge for organizations in highly competitive sectors, particularly banking, where retaining existing clients is more effective than acquiring new ones. This study develops a machine learning–based system for churn prediction in the banking sector. The system was designed in Python, integrating modules for data preprocessing, feature engineering, and model training and evaluation. Within this system, five models are trained using five different algorithms (one training algorithm produces one trained model): Logistic Regression, Linear Discriminant Analysis (LDA), Decision Tree, Random Forest, and Gradient Boosting, based on an initial dataset of 10,000 clients, stratified prior to use, obtained from a public Kaggle dataset. For the evaluation of model performance, metrics such as accuracy in churner identification, precision, recall, and F1-score were employed, along with supplementary procedures such as ROC–AUC curve analysis and bootstrapping for model stability assessment. Results demonstrate that ensemble models, particularly those trained using the Random Forest and Gradient Boosting algorithms, outperform baseline approaches across all selected evaluation metrics. Moreover, the 95% confidence intervals, which are narrower than those of the other trained models, along with the ROC–AUC curve analysis, indicate stable predictive performance for these two models. As a result, these algorithms are recommended for identifying potential churners across various business and industrial use cases. Their reliable performance indicates that they can effectively support customer retention campaigns.
Agronomy students need to identify agricultural pests like fruit flies of the genus Anastrepha, a task often hindered by the lack of interactive tools. This study develops a mobile "virtual laboratory" that integrates augmented reality (AR) with deep learning for species recognition in biointensive fruit-growing ecosystems. The MobileNetV2 architecture was chosen over VGG16 for its greater efficiency-to-accuracy ratio. The Barracuda inference worked with the Unity engine so that it could handle data in real time. A test set of 200 images was used for experimentation, yielding an F1 score of 0.98 and an accuracy of 98%. Usability tests were also conducted to measure the optimal AR scanning distance, which showed that the ideal distance was 15cm. Therefore, it can be concluded that this is an effective tool for species identification, with low latency and no internet connection required for cloud services.
Dynamic environments and coordinated malicious behavior remain major challenges in virtually any Internet of Vehicle system, making trust a challenging proposition. This paper devises a novel approach to achieve blockchain consensus through collective computation and dynamic Work evaluation to detect/evade Roadside station collusion. It creates Trust evaluation based on historical reviews, behavior significance, and scoring - variation and plugs into adaptive difficulty adjustment. This leads to differences in Shapley value, and similarity-based metrics, allowing the system to identify colluding RSUs and influence their mining probability. Nash stability and compatible desires are guaranteed by game-theory analysis. Planned framework is seen to enhance packet delivery ratio up to 17.4%, lessen end-to-end delay up to 21.6% and increase throughput by 13.9% when compared with basic planned frameworks as shown in reproduction results. This framework provides a scalable and reliable solution to secure and fair data sharing in blockchain-based Internet of Vehicles.
The aim of this study is to empirically assess whether the regular use of AI in the educational process can be considered as an indicator of the foster of digital competencies relevant for the future labor market and innovative culture. The theoretical basis is human capital theory, within which the use of AI is interpreted as a form of early investment in digital competencies. The empirical base includes survey data from students in Kazakhstan, Poland, and Ukraine. To test the research hypothesis a parametric t-test with a threshold of 50% is used. The findings indicate that students can act as drivers of AI innovation only if artificial intelligence is widely and constantly integrated into educational practices. From a practical perspective, the results highlight the importance of educational policies focused on the development of digital competencies as a key element in shaping the innovative potential of the future labor market.
Fraud in health insurance leads to huge financial losses and claim processing systems inefficiencies. In this paper, the proposed system to be used is a hybrid Genetic Support Vector Machine (GSVM) model of a Decision Support System that will be effective in detecting fraud. FIC-GSVM method combines Genetic Algorithms to select the most optimal features and Support Vector Machines to classify features. The model is tested with the National Health Insurance Scheme (NHIS) data on Ghana, which is a set of structured and coded healthcare claims. Experimental findings indicate that the proposed approach has a 90.5% accuracy, as well as high precision, recall, and reduced processing time as compared to the current approaches. The system facilitates automated, scalable, and near real-time fraud detection, which can be an effective solution to healthcare insurance analytics.
Transportation planning is about coming up with and looking at different policies, programs, and projects that affect how people and goods get around in a certain area. In this field, technicians rely on quantitative methods and modeling tools to carefully examine the effects and the real usefulness of the measures proposed to improve the transportation system. To do this, mobility surveys are often used to evaluate the level of acceptance/preference of users for new transport services or improvement of existing ones (e.g., redevelopment of a transport terminal; new bus connections/stops; new services for travelers). Often in these surveys, hypothetical scenarios of services/facilities that do not yet exist are submitted to users through the so-called Stated-Preferences (SP) surveys. Although there is a copious literature on this type of survey and on the models and methods that can be derived from this data, little attention has yet been paid to the quality with which the hypothetical scenarios are presented to travelers (e.g. images, videos or simple text description) and above all to their real ability to identify themselves with these hypothetical contexts (e.g. the choice between a new and non-existing train service competing with the usual bus service used every day). This issue is central for mobility planners, because a low capacity of users’ identification corresponds to a low quality of the models and estimates that derive from them, and which can limit/undermine the entire planning process. The recent disruptive development of immersive technologies (es, virtual reality; augmented reality), developed mainly for other sectors (e.g., gaming and entertainment; education and training; retail and e-commerce; architectural design and visualization) today also find a possible application in the transport sector as a usufull tool for evaluating mobility policy effectiveness within, for example, SP survey. This paper aims to look at the main immersive technologies, like Virtual Reality (VR), Augmented Reality (AR), and Mixed Reality (MR), and see how they are used in transportation planning. It also tries to point out what works well and what doesn’t for each technology. The results show that extended reality technologies (VR, AR, MR) make transport SP surveys feel more realistic and help users relate better to the scenarios, which in turn makes transport models more reliable. VR gives a fully immersive experience but can be expensive and tricky to use, while AR and MR offer a mix of realism and easier access. We also need standard methods and long-term studies to really understand how these technologies affect planning decisions.
Smart campus environments increasingly rely on IoT platforms to provision, configure, monitor, and control devices across multiple building spaces. As deployments scale, differences in communication protocols, payload structures, and data models complicate integration, monitoring, and visualization. This paper proposes a reusable room-centric configuration approach for IoT device management and monitoring, implemented in Node-RED as a visual orchestration layer. The technical contribution is a template-based method that transforms heterogeneous MQTT device payloads into unified room-level data objects reusable across similar spaces with minimal reconfiguration. The approach is validated through smart campus scenarios in a Smart Lecture Hall, covering network connectivity, energy consumption, and electricity availability. To evaluate deployment efficiency, a simulation experiment based on a real configuration session was conducted. Results show that reusing the initial room setup as a Node-RED template reduced configuration time for a five-room deployment from 70.47 to 35.47 minutes, improving the overall process by 49.66%.
Nowadays, three main approaches are used to manage projects - traditional, agile, and hybrid (or combining the previous two). Traditional project management methods involve the entire project life cycle in several phases, where the next phase can only start if the previous phase is completed correctly. The customer then receives a near-final product, which can be costly to change. At the same time, the customer also needs a complete picture of the final product. Agile methodologies are based on a regular inflow of products at regular intervals. This allows the customer to provide feedback during its development. Choosing the appropriate project management style can be difficult, as there are many criteria to consider that influence this decision. This article discusses creating an application to help support a project manager's decision on which project management style to choose - agile or waterfall, or how much to involve agile elements. The application uses a previously built ANP model based on the PRINCE2 ® Agile standard, the Agile Manifesto, and the SCRUM framework. This work presents a pilot version of the decision-support tool. The underlying ANP model and application have not yet been validated against real project outcomes; their calibration and validation using historical data and expert assessment are left for future research.
This paper presents a fully analytic, neuro-symbolic forecasting pipeline that automatically extracts symbolic rules from multivariate time-series and embeds them as differentiable constraints in an Extreme Learning Machine (ELM) predictor. An ELM-based auto-encoder (ELM-AE) first learns latent descriptors in closed form; these descriptors are discretised and mined for frequent temporal patterns. Association-rule and inductive-logic discovery yield a library of Answer Set Programming (ASP) clauses. The clauses are re-encoded as hinge-loss penalties inside a second closed-form ELM forecaster (ELM-F), yielding long-horizon predictions that are both accurate and rule-consistent. The entire system trains rapidly on commodity hardware and runs in real time on microcontrollers, making it attractive for safety-critical digital-twin applications such as those for battery management systems. Beyond its novel algorithmic contributions, this paper also provides a hands-on tutorial on applying analytic ELM methods and symbolic rule integration to practical forecasting problems.
This paper introduces Gamify Away, a tool currently under development designed to support brainstorming sessions for the creation of game-based learning (GBL) experiences. Developed within the context of a doctoral research project, the tool aims to unlock the initial stages of GBL design, enabling participants to produce a concise game concept that can serve as the foundation for subsequent development phases. The results obtained from its experimentation indicate high levels of satisfaction regarding the tool’s effectiveness, particularly in terms of defining learning outcomes to be obtained, participant engagement, and its capacity to stimulate idea generation. Participants also expressed strong recommendations for its use in future brainstorming sessions. Although further iterations are still required, Gamify Away is believed to be applicable across various contexts and domains, helping to enhance initial approaches to creating GBL experiences.
External audits are making better using AI technologies for boosting efficiency, accuracy, and how quickly risks are spotted. In fact, AI helps make the audit process more reliable and effective by taking over repetitive tasks, strengthening predictive analysis, and improving fraud detection. This study aims to determine how AI techniques, like expert systems, artificial neural networks, and robotics, are improving audit quality in Jordanian audit fi rms. The study population consists of all practicing external auditors in Jordan, totaling 542 auditors. A sample of 244 auditors was used to achieve the study's objectives. The study results showed that AI technologies positively impact enhancing the quality of external auditing.
This study examines the effects of capital structure on the profitability of 100 IDX-listed firms from 2004 to 2014. This research aims to disentangle the relationships between capital structure (DER) and profitability (ROA, ROE, and NPM) using the Hausman panel-data test and the Fixed Effects and Random Effects Models. The ROA and ROE ratios suggest that some financial leverage enhances asset efficiency. In contrast, the relationships between NPM and DER suggest positive net interest income, implying that interest expenses are offset by funds borrowed from financial institutions. This study examines the effect of some control variables on the relationships described above. The interest and inflation variables are the control variables of moderate capital structure efficiency and NPM, but only weakly and negatively. In addition, the contextual variable was found to moderate the relationships positively.
The blending of econometric evaluation techniques with big data analytics and dynamic modeling forms the core of developing prescriptions for adaptive, unprecedented, environmentally sustainable approaches to the management of the environment and natural resources. The model enables the design and assessment of policies and measures that have not been subjected to field surveys but remain highly relevant and impactful. Research creates big data 'snapshots' of chains of phenomena across space and time, including, but not limited to, forest cover, water quality, social activity, per capita CO2 emissions, and geo-economics. Dynamic systems modeling sets the system's boundaries, identifies the primary points of intervention for optimal environmental and socio-economic sustainability, and assesses the nonlinear causal system of the policy model within those boundaries. Econometrics focuses on the multivariate regression of integration and structural disintegration of the policy and institutional framework, the fiscal ecological cost of nested ecological units.
Structural Equation Modeling (SEM) is widely used to analyze relationships among latent and observed variables in social and behavioral research. However, researchers face methodological, statistical, and practical challenges that threaten model validity and interpretability. This study conducts a systematic literature review of 35 peer-reviewed articles (2020–2025) using the PRISMA framework. The findings identify five major challenges: model misspecification, measurement-model instability, data quality and sample-size limitations, statistical and estimation issues, and practical implementation constraints. Common problems include overfitting, under-identification, low factor loadings, non-normality, and convergence failures, which are linked to violations of identification conditions, distributional assumptions, and finite-sample properties. Software heterogeneity and incomplete reporting further limit replicability. The results emphasize that robust SEM inference depends on strong theoretical grounding, appropriate statistical assumptions, and transparent reporting practices.
In today’s increasingly interconnected world, organizations interact with complex data ecosystems that feature multiple independent stakeholders. Establishing trust, transparency, as well as integrity concerning such ecosystems is a difficult task, particularly if stakeholders feature conflicting interests with different, unintegrated information systems. Conventional, centrally managed systems are prone to single-point failures, manipulations, as well as delays within the verification process. Blockchain technology, with its decentralized, immutable system, provides a sound basis for addressing such issues. BC-4Track, a blockchain-enabled system, is presented as a solution that provides secure, transparent, as well as tamper-proof management of lifecycle information within the automotive industry. This system incorporates IoT-based smart kits for real-time information acquisition, Ethereum-based smart contracts for automated record management, as well as a web-based decentralized application (DApp) that enables stakeholder interaction. Thorough analyses were carried out with a focus on system functionality, system efficiency, and system reliability. System usability, system responsiveness, and system security were validated. These results demonstrate the applicability of blockchain-based solutions in such environments with multiple stakeholders, with regard to issues such as trust.
This study investigates the influence of digital technology on the economies of developing countries. It examines the Digital Adoption Index, ICT expenditures, Ease of Doing Business Index, and Gross National Income per capita to assess their relationship with economic performance. Using endogenous growth theory, the study explores how technological progress, business environment, and income levels contribute to economic growth and support Sustainable Development Goals 8 and 9. Panel data from 25 developing countries (2010–2022) obtained from the World Bank are analyzed using a fixed effects regression model, with unit root and cointegration tests ensuring robustness. Results show that higher digital adoption, ICT spending, and improved business conditions positively impact economic performance and GDP per capita. The findings emphasize the importance of policies that promote technology adoption and regulatory improvements to achieve sustainable economic growth and digital transformation aligned with global development goals.
The objective of this paper is to tackle the problem of online state estimation for supercapacitors by using a strong high-gain observer. To monitor and control in real-time systems in the renewable energy domain, such as energy storage systems and electric vehicles, it is essential to accurately estimate supercapacitor parameters. Traditional estimation methods can be computationally expensive or sensitive to measurement noise. Presently, we propose a high-gain observer designed under the assumption that the supercapacitor parameters (series resistance R_s and capacitance C) are known and constant. Assuming system observability, the observer ensures asymptotic convergence of state estimates. The simulation results demonstrate fast convergence, less than 500 seconds of simulation, and robustness to measurement noise (Maximum relative error of 1%), confirming the suitability of the proposed approach for real-time implementation.
Higher education's digital transformation is a current trend to meet the demands of the digital economy and plays a vital part in modernizing the educational system. However, there are many obstacles in the way of higher education institutions' digital transformation. Higher education institutions' digital transformation initiatives must be successfully developed, categorized, and assessed in order to be successful. In addition to offering competitive advantages to Vietnamese universities today, this article analyzes the strategies employed in the process of digital transformation in higher education institutions across the globe and suggests the development of some suitable digital transformation models.
