Volume 8.
Issue 05.

Clean cooking adoption remains a persistent challenge in low- and middle-income urban contexts despite technological advancements and policy interventions. This study develops a socio-behavioral framework to explain sustainable adoption patterns of clean cooking technologies in urban Kigali households. Drawing on ethnographic insights, the research integrates behavioral, economic, and socio-cultural determinants influencing adoption decisions and long-term usage. The study critically examines the limitations of conventional diffusion models, particularly the energy ladder hypothesis, and proposes a multidimensional framework that accounts for trust, time-use practices, culinary preferences, and institutional dynamics. The findings reveal that adoption is not a linear transition but a complex, iterative process shaped by contextual realities and behavioral inertia. The proposed framework advances theoretical understanding and provides actionable insights for policymakers and practitioners aiming to achieve durable clean cooking transitions.

This work considers the issue of integrating Internet of Things (IoT) resources into business process management systems. The modern business environment requires processes to be flexible, reusable, and quickly adapt to different conditions. Therefore, configurable process models are considered an effective approach that reduces the need to develop processes from scratch and forms options that meet business requirements. However, the heterogeneity of IoT devices, resource limitations, energy consumption, computational costs, and interoperability problems complicate their full integration into business processes. Existing studies have mainly analyzed the control flow or human resources, and the perspective of IoT resources has not been sufficiently studied. This work proposes a formal introduction of IoT resources into business process models, taking into account their sharing and redeployment properties, and a configurable distribution approach. The proposed model is evaluated based on real data, and its effectiveness and practical applicability are substantiated.

The study presents a comprehensive examination of microbiological threats associated with the formation of microbial biofilms on the working surfaces of medical instruments and equipment in the healthcare realities of 2024–2025. On the basis of updated information from international organizations, as well as a synthesis of specialized publications indexed in Scopus and Web of Science, the analysis elucidates the problem of heightened resistance of biofilm communities to standard disinfection and sterilization regimens. Key strategies by which microorganisms maintain viability within the extracellular polymeric matrix are described in detail, including mechanical protection, restricted diffusion of antimicrobial factors, and the development of heterogeneous microzones with altered metabolic activity. In parallel, the dynamics and statistical regularities of healthcare-associated infections (HAIs) at the global level are examined, enabling biofilm-associated persistence mechanisms to be linked to epidemiological trends. A separate emphasis is placed on the transformation of international regulation and practice-oriented requirements for the reprocessing of medical devices, including the release of updated standards in the ISO 15883:2024/2025 series and their significance for evidence-based validation of technological processes in central sterilization departments. Within a comparative analysis, contemporary approaches to detecting biological contamination are considered, including ATP bioluminescence and fluorescent visualization methods, which are characterized by differing sensitivity and applied diagnostic value at stages of cleanliness control. The effectiveness of innovative enzymatic detergents in acting on mature biofilms is additionally assessed, where the critical factor is the capacity to degrade matrix components and thereby increase the accessibility of microbial cells to standard decontamination interventions. The integrated conclusions underscore the fundamental importance of transitioning from universal control schemes to biofilm-oriented validation protocols aimed at reducing residual contamination, enhancing patient safety, and limiting the contribution of the clinical-diagnostic environment to the accelerated spread of antimicrobial resistance.

Entrepreneurial skills development programme design for youths have come among other things, part of the policy thrust of the Nigeria Government in order to promote self-employability and reducing high unemployment rate among youths. This study was conducted to determine the difference among youths who had sexuality education knowledge before acquiring entrepreneurial skill and youths who did not have sexuality education but acquire entrepreneurial skill in Nigeria. To guide the study, two objectives, two research questions and two null hypotheses were formulated. The descriptive survey research design was used for the study. The population of the study comprise 151 million youths in Nigeria, 400 youths was obtained based on Taro Yamene Formula then multiply by 2 to have 800 youths for more generalization of the result of the Finding. Using accidental sampling technique. The research questionnaire designed and used for the study were subjected to reliability test on respondents and the data collected for the reliability were subjected to Cronbach Alpha reliability test, which yielded a coefficient of 0.80. Findings of the study show that there is different among youth who had creative and critical thinking skill based on sexuality education knowledge before acquiring entrepreneurial skill with youths who had not but acquired entrepreneurial skill. It is recommended that sexuality education should be introduce as a compulsory subject to study in both secondary and tertiary institutions to enable youths acquire salient information and skills on human sexuality and reproductive health

The fast usage of generative artificial intelligence (AI) in professional services Customer Relationship Management (CRM) systems has increased opportunities for hyper-personalized engagement with clients, but such change creates new risks in regulated practices such as legal services, where client data isn’t just sensitive, it’s protected by attorney–client privilege and a thicket of compliance rules like GDPR and CCPA. Recent breaches of law firms' confidential client files owing to breaches of data privacy have already shown how messy this can get: the Proskauer Rose breach in 2023, for example, exposed sensitive deal documents, and Bryan Cave Leighton Paisner faced a similar crisis in 2024. Cases like these indicate the urgency for a generative AI framework that preserves privacy while synergistically maximizing the benefits of enhanced personalization.

This paper presents a privacy-preserving generative AI framework that is designed specifically for legal CRM scenarios. This idea is a multi-layered framework approach, which is differential privacy baked into the data, federated training so information doesn’t have to leave its source, compliance checkpoints to catch GDPR/CCPA gaps, and audit trails that hold systems accountable. A synthetic set of anonymized legal CRM records were produced to test the application of the framework. The results showed a 59% reduction in the exposure to privacy risk, a 40% improvement in compliance scores, three times more audibility, and acceptable levels of personalization relevance. In addition to the quantitative results, expert validation from legal technologists and compliance specialists for the adoption of frameworks/case study's in practice was obtained.

In summary, this research offers three contributions: (1) this is the first research to focused on generating AI-driven personalization aligned to compliance-driven privacy safeguards for legal CRM; (2) this study offered a hybrid evaluation process that combines synthetic benchmarks with expert input for evaluation of adoption; and (3) this study contributes to shifting the conversation away from maximum personalization, irrespective of regulations/standards and towards transparency, trust, and compliant, future proofed practices in a regulated domain of legal CRM.

This paper discusses the application of parallel computing technologies in artificial intelligence and machine learning processes. The study focuses on heterogeneous computing systems based on CPUs and GPUs, as well as the use of CUDA technology for parallel data processing. Experimental results show that GPU-based parallelization significantly improves computational speed and reduces execution time compared to traditional CPU-based processing. The research confirms the effectiveness of GPUs in accelerating machine learning algorithms and other computationally intensive tasks.

E-commerce platforms generate massive volumes of user-generated product reviews, making sentiment-aware recommendation systems essential for improving personalization and decision-making. This study aims to develop a high-performance real-time product recommendation framework by integrating sentiment analysis with deep learning techniques. The proposed method utilizes Amazon review data, which is preprocessed through feature extraction using TF-IDF, followed by class balancing using SMOTE. Accurate sentiment categorization in user evaluations is achieved by using a stacked LSTM-based DL model that captures contextual and sequential relationships. Accuracy, Precision, Recall, F1-score, and AUC-ROC measures are used for model evaluation after training using Binary Cross-Entropy loss and the Adam optimizer. Based on the experimental findings, the suggested model regularly beats more conventional ML models like Decision Tree, Logistic Regression, and Naïve Bayes, as well as other DL methods like RNN and DeepFM, with an accuracy rate of 98.42%. This study primarily contributes by enhancing the relevance of recommendations in real-time via the merging of balanced learning and sentiment-driven LSTM modeling. In conclusion, the framework provides a scalable, accurate, and robust solution for large-scale deployment in modern e-commerce systems.

The exponential rise in global online transactions intensified the complexity and frequency of identity-fraud attacks, necessitating distributed and intelligent verification systems. A hybrid multi-cloud identity verification framework was designed and evaluated by combining Amazon Web Services (AWS) and Google Cloud Platform (GCP). The framework integrated event-driven serverless components, cross-cloud streaming analytics, and machine-learning-based anomaly detection. In controlled experiments spanning 500–5000 requests per second, P95 decision latency was reduced by 28–31% relative to single-cloud baselines (290ms vs. 420/405ms), and the false-positive rate was reduced by 33–38% (2.8% vs. 4.5%/4.2%). Overall detection accuracy reached 96.4% and end-to-end system availability reached 99.9%.

This article examines the scientific, theoretical and practical significance of using innovative educational technologies in the process of developing students’ financial literacy. It discusses the essence of financial literacy, modern requirements for it in the context of the digital economy, and the pedagogical potential of such technologies as generative artificial intelligence, adaptive learning systems, simulation methods, microlearning, digital assessment and learning analytics. The results of the study show that innovative technologies serve as an effective tool for developing students’ skills in planning personal and family budgets, controlling expenses, forming savings, understanding credit and debt obligations, using digital payments safely, and making informed financial decisions.