Asian Journal of Research in Computer Science
Not a member yet
792 research outputs found
Sort by
The Role of Data Science in Improving Healthcare Access and Equity
No full textThe integration of data science in healthcare has transformed the landscape of medical decision-making, resource allocation, and patient care. Using big data, electronic health records (EHRs), and social determinants of health (SDOH), data science offers innovative solutions to identify healthcare disparities, optimize interventions, and enhance patient outcomes. Geospatial analytics and predictive modeling have proven effective in mapping underserved regions and forecasting disease trends, thereby enabling targeted policy interventions. However, challenges such as algorithmic bias, data interoperability, and privacy concerns remain significant barriers to widespread adoption. Ethical considerations, including fairness in AI-driven healthcare models, require urgent attention to ensure that data-driven interventions benefit all populations, especially marginalized communities. This paper explores the role of data science in improving healthcare access and equity, emphasizing predictive analytics, artificial intelligence, and machine learning applications. The study highlights the necessity of diverse and representative datasets to mitigate biases in predictive models and promote equitable healthcare delivery. Furthermore, the implementation of fairness-aware AI techniques can help prevent discriminatory outcomes and improve trust in data science applications. By addressing these challenges, data science has the potential to bridge gaps in healthcare access, ensuring that technological advancements translate into meaningful improvements in public health. The findings reveal the importance of collaboration between policymakers, healthcare providers, and data scientists to maximize the benefits of data-driven healthcare. This paper advocates for a systematic approach to integrating data science methodologies into healthcare policies to create a more inclusive and effective healthcare system
The Evolution of Consumer Trust in E-Commerce: Exploring Digital Strategies for Enhanced Loyalty
No full textThis review titled (The Evolution of Consumer Trust in E-Commerce: As evidenced in the article entitled Consumer Trust Dynamics in the Era of Digital Technologies: Uncovering Digital Approaches for Improving Loyalty, the development of consumer trust in e-commerce has emerged as a key influence in the new generation of fragmented business web environments. The research question of this paper is to establish how the use of digital tactics impacts consumer trust and improves loyalty in the growing market of e-commerce industry. The main issue stems from the high level of doubt from the consumers about the safety of their important information and personal data, fraud that affects loyalty towards the online purchase. This study aims at examining various digital communication strategies with the aim of determining which of them facilitates the building of trust; This is achieved through examining transparency in communication, proper addressing of user needs, adequate security measures and efficient management of customers’ feedback. Methodology My review is researching 30 articles in the period of: (2019 - 2025). Reflecting on the current literature and shopping trends over the last five years, this paper reveals that digital innovation like automation through Artificial Intelligence, blockchain technology, and personalized marketing for a customer have minimized the breach of trust and have encouraged long-term customer relationships. The results imply that, apart from trust, ethical and consumer-oriented strategies enhance the level of customer loyalty. Last but not the least, this research outlines a roadmap for e-commerce organizations to construct a long-lasting ecosystem that is based on trust and reliability
Empirical Analysis of the Impact of Routing Protocols on Malicious Node Detection in Opportunistic Networks
No full textThe Internet of Things (IoT) has revolutionized how devices communicate and interact, enhancing the effectiveness of diverse applications. Opportunistic IoT (O-IoT) networks, characterized by dynamic, decentralized, and resource-constrained architectures, are increasingly being adopted in environments with unstable or absent infrastructure, such as disaster-stricken areas. However, the transient connectivity and mobility in these networks pose significant security challenges, particularly in detecting malicious nodes. This paper investigates the influence of routing protocols—Epidemic, Spray and Wait, and ProPHET—on malicious node detection efficiency in O-IoT networks. The problem is compounded by varying parameters such as buffer size and Time-to-Live (TTL), which affect both performance and security. We used the ONE simulator to build a simulation-based framework that examined different routing protocols with buffer sizes ranging from 5 to 20 MB and TTL values between 100 and 400 seconds. The results demonstrate that routing protocol selection significantly influences detection capabilities and resource consumption. Epidemic routing provides high delivery probability (80-85%) and packet delivery (85%), but incurs substantial overhead (70-80%) and latency (70-75%). Spray and Wait reduces overhead (40-55%) at the cost of lower delivery rates (60-65%). ProPHET achieves the best balance, maintaining moderate delivery rates (70%) while minimizing overhead (30-40%), buffer time (40%), and latency (40%). Our findings provide valuable insights for designing secure and efficient O-IoT systems, especially in resource-constrained environments such as disaster management and military applications. The results highlight the importance of protocol selection and parameter tuning in achieving optimal detection efficiency while maintaining acceptable network performance. Future work will focus on enhancing security measures through advanced cryptographic techniques and machine learning integration to improve malicious node detection without compromising network performance
A Novel AI-Driven Homomorphic Encryption Framework for Secure Real-Time Telehealth Data Analysis
No full textEnsuring privacy in AI-driven telehealth analytics remains a persistent challenge, as conventional cryptographic methods struggle to meet real-time and compliance requirements. This research developed and validated an AI-driven homomorphic encryption framework for secure real-time telehealth data analysis, addressing critical privacy challenges in medical IoT systems. The study designed a proactive threat intelligence system, developed a predictive analytics framework, and guided secure implementation. A review of existing cryptographic solutions identified gaps in scalability and real-time processing. Using a quantitative experimental design, synthetic telehealth datasets, hybrid CKKS-BFV schemes, and neural network optimization were employed. Implementation in Python with SEAL and TensorFlow was tested across computational, security, and compliance metrics. Results showed a 23.7% overhead reduction, sub-535 ms latency for 5,000 records/sec, and 96.9% HIPAA compliance, with attack success rates below 6%. Synthetic data achieved 99.3% quality, and performance improvements over AES-256 and Paillier were statistically significant (p < 0.001). The hybrid scheme outperformed single approaches by 18.4%, supporting scalable, accurate analytics. Despite synthetic data limitations, findings confirm the framework’s ability to secure telehealth data and enhance clinical decision-making. Future work includes real-world dataset development, explainable AI integration, clinical deployments, and adaptive algorithms for emerging threats
Rethinking Data Architectures in the Face of Information Diversity and Exponential Growth
No full textSubject: The subject of this article is the analysis of the impact of exponential growth in data volume (up to petabytes and exabytes) and variety (Big Data) on data management architectures and methodologies.
Aims: The objective is to identify the challenges in processing and integrating large volumes of heterogeneous data and to conduct a comparative analysis of modern approaches.
Methodology: The methodology employs systematization, generalization, and comparative analysis of architectures (NoSQL, Data Lake, Hadoop, Spark, Flink) and methodologies (Agile, DevOps, Data Governance, Data Mesh, Data Fabric).
Results: This manuscript focuses on a pivotal topic in Big Data management, exploring the interplay between data growth, architectures, and methodologies. Results indicate that traditional relational DBMS (Database Management Systems) exhibit significant limitations in horizontal scalability and unstructured data processing, whereas NoSQL solutions (document, columnar, etc.) offer the schema flexibility and scalability required for Big Data. Distributed systems, such as Spark and Flink, provide orders of magnitude higher performance for analytical and streaming tasks compared to traditional approaches. The study underscores the critical interconnection between architecture selection (e.g., Data Lake for flexibility) and methodology adaptation (e.g., DataOps for speed, Data Governance for quality control) for effective data integration and management. The scope of application includes the design of data management systems and the selection of optimal technology combinations (e.g., ELT instead of ETL in Data Lakes) for analytics. Its systematic comparison of key technologies and frameworks addresses a gap in literature that often treats these elements separately. Real-world case studies enhance practical relevance, offering valuable guidance for practitioners. It contributes meaningfully to the scientific community by synthesizing selection criteria for effective Big Data systems. A conclusion is drawn regarding the necessity of an integrated approach that combines horizontally scalable architectures, modern processing tools, and flexible yet governed methodologies for successfully handling Big Data
Blockchain-based Supply Management System for Enhancing Transparency and Accountability
No full textAs traditional supply chain operations continue to pose challenges to the businesses, the adoption of Blockchain to solve issues of traceability and integrity in global supply and logistics networks is increasingly evident and persuasive across multiple geographies. The objective of this research study was to build a Blockchain Optimized Chain of Accountability (BOCA) to eliminate delays, inaccurate inventories and vulnerability to fraud; thereby enhancing transparency, accountability and efficient supply chain ecosystem. The methodology used is the Structured System Analysis and Design methodology to allow necessary changes, and adjustment to the system to be made quickly; resulting in a more robust, and user-friendly interface that aligned with users’ needs and expectations. The system multi-tier architecture makes transactions visible to authorized users and the automatic trigger of a smart contract once a product moves from one party to another further offers secure and immutable transaction. The programming languages and tools used for the coding and development of the system are JavaScript (Vue.Js) and HTML/CSS for the user interfaces; JavaScript (Node.Js) for the blockchain; Go for development of blockchain; and SQL for data storage. The innovation will enable companies, businesses or individuals to order goods and raw materials for production and keep track of their inventory, make secure payments, gather, analyze and maintain tamper-proof record of supply chain events. By integrating smart contracts and real-time tracking, the system will enhance transparency and trust among stakeholders and largely benefit companies aiming at modernizing their supply chain logistics networks across the globe. The paper thus offers valuable insights for academics, practitioners, and policymakers to refine and advance blockchain-based logistics solutions in the global supply chain ecosystems
EngageNet: A Model for Evaluating Student Engagement through Facial Expression and Behavior Analysis
No full textOnline learning has emerged as a prominent trend in modern education, driven by its flexibility, accessibility, and capacity to support personalized learning experiences. However, despite these advantages, one of the most pressing challenges it faces lies in maintaining and accurately evaluating the quality of teaching and learning. A particularly critical aspect is the assessment of learner engagement in virtual environments. In fact, traditional approaches to assessing student engagement, which depend on synchronous, face-to-face interaction, frequently prove inadequate in virtual learning environments where such real-time communication between educators and learners is restricted. Therefore, this study introduces a deep learning-based model that combines facial emotion recognition, gaze direction tracking, and eye openness analysis. By integrating these emotional and behavioral characteristics, the model offers a comprehensive and objective approach to assessing learner’s attention throughout online instruction. To support the development and validation of this model, a specialized dataset was proposed, capturing a diverse range of engagement scenarios. Experimental evaluations demonstrate that the proposed method achieves a notable accuracy of 79.76%, underscoring its effectiveness and robustness in capturing learner engagement dynamics. These findings suggest that the model holds strong potential for enhancing the monitoring and personalization of online learning experiences, thereby contributing to improved educational outcomes in virtual classrooms
Machine Learning-based Customer Churn Analysis in Telecommunications Using Support Vector Machines
No full textFaced with globalization and increasing competition, the information available via the Internet and the many connected objects continues to increase. This explosion of data, often heterogeneous and from diverse sources, poses major challenges in terms of storage, analysis and exploitation. This paper is the result of the present research on the analysis and classification of churning customers in a telecommunications company. These data, often heterogeneous and coming from various sources, require in-depth analysis as well as new storage and exploration paradigms to extract value from them. The dataset used for the implementation of the prediction model is based on the existing reality, within the telecommunications company named Airtel Congo; on the customer management policy, more precisely the customers who are candidates for churn.
In the telecommunications sector, companies accumulate large amounts of information about their customers, coming from multiple sources: social networks, telephone platforms, electronic messaging, open data, geolocation, and many others. The intelligent exploitation of this data allows to better understand user behavior and anticipate key phenomena, such as “ churn ” – i.e. customer unsubscription.
Churn is a major strategic issue for telecommunications companies, as customer loss leads to high costs related to new subscriber acquisition and reduced revenue. Thus, identifying customers at risk of churn and understanding the underlying factors are essential to implement preventive actions and build customer loyalty.
In this study, a machine learning model based on support vector machines (SVM) was proposed to analyze and classify churning customers. This algorithm, recognized for its ability to handle complex and multidimensional data, is implemented using the LIBSVM library in the C# language. The objective is to build a powerful predictive model to identify, with high accuracy, customers likely to leave the operator, in order to optimize retention strategies and maximize customer satisfaction. Based on various techniques such as supervised and unsupervised learning, it allows to discover hidden patterns and make accurate predictions. SVM, in particular, illustrate the effectiveness of supervised approaches, by allowing an optimal separation of classes through the maximization of the margin
Integrating AI for Sustainable Supply Chain Optimisation in Ethical Fashion: A Case Study
No full textIn the recent past, growth in AI has been key to providing chances to enhance supply chains, proven to be efficient economically as well as environmentally. Some of the most promising AI use cases include demand forecasting, inventory management, and supplier validation to support the brand\u27s decision-making processes. The present study will address the state and directions of the implementation of AI technologies into the sustainable supply chains of the ethical fashion industry in order to determine more efficient and low-cost measures to minimise the deleterious effects on the environment whilst increasing the overall productivity. Thus, the research aimed to show how the concept of AI can be effective in addressing the issues in sustainable fashion brands, which include textile waste management, ethical sourcing, and increasing consumer trust in the industry by making the companies’ supply chain transparent and responsible. This research adopted a qualitative case study research approach. This research focused on a mid–sized ethical fashion brand in Europe that follows a sustainable and fair labour policy. The following methods were used in the research: A survey of the supply chain indicators, interviews with operational staff, assessment of AI tools that are used in demand forecasting and validation of materials. The study showed how the use of demand forecasting tools made using Artificial Intelligence decreased overproduction by 30%, as compared to an increase in productivity through seasonal planning and inventory control. This system also enabled the company to ensure compliance with over 95% of the suppliers. Additionally, lead time was reduced by about 18%. In addition, the system was helpful to identify and track sources of raw materials such as organic, naturally-sourced cotton and recycled fabric for assessing the suppliers’ ecological and workers’ treatment performance. This not only increased the supply chain and product identity revelation but also aided to adapt with the changing customer perception as to how they want their products to be made. The work presented in this paper shows that non-tech maintenance companies can also seize the opportunity and augment their sustainability performance with the help of the scalable AI solutions. Therefore, it is important to note that with the principles cited in this research, AI in ethical fashion is set to revolutionise the sector through enhanced, efficient supply chains, supply chain sustainability, as well as ethical fashion consumer loyalty. 
A Systematic Review of Privacy-preserving Techniques in Databases
No full textAims: This systematic review aims to explore how artificial intelligence (AI) enhances privacy-preserving techniques in database systems, focusing on anonymization, differential privacy, and secure multi-party computation (SMPC), while evaluating their effectiveness in balancing privacy and data utility and identifying implementation challenges.
Methodology: A comprehensive search strategy was applied using predefined search strings targeting AI-driven anonymization, differential privacy, and SMPC in database systems. The initial search yielded 62 records, which were screened based on inclusion criteria (peer-reviewed studies published in English between 2020 and 2025, focusing on AI-enhanced privacy-preserving techniques in databases) and exclusion criteria (non-peer-reviewed sources, studies lacking empirical results or database focus). After screening and eligibility assessment, 20 studies were included. Data extraction focused on sub-themes, AI enhancements, application domains, challenges, and effectiveness metrics, followed by qualitative thematic synthesis to address the research questions.
Results: Of the 20 included studies, AI-driven anonymization reduced information loss by up to 12% in accuracy improvements using blockchain schemes and lowered execution times, while clustering methods enhanced privacy in social networks. Differential privacy preserved 60.81% data originality while reducing privacy risks by 20.05% in hybrid models. SMPC enabled secure genomic data exploration, with fast Machine learning training (<45 seconds for binary classifiers), and processed 10,000 variables across 20 parties in under 5 minutes using no-code tools. Challenges included scalability issues and privacy-utility trade-offs like excessive noise in biomedical databases.
Conclusion: AI significantly enhances privacy-preserving techniques in databases, enabling effective privacy protection with practical utility across healthcare and social networks. However, challenges like scalability and privacy-utility trade-offs highlight the need for future research into combined methods and standardized evaluation frameworks to ensure reliable, widespread adoption in database systems