Proceedings of International Conference for ICT (ICICT) - Zambia

A Gis-Based Mobile Application for Real-Time Disease Outbreak Monitoring and Prediction in Zambia

2025-12-25T08:52:55+00:00

Timely and effective response to disease outbreaks remains a persistent challenge, especially in resource-constrained environments such as Zambia. This study presents the development of a GIS-based mobile application that combines Geographic Information Systems (GIS) with artificial intelligence (AI) to enhance real-time outbreak monitoring and decision-making. The system integrates geospatial visualization with AI-driven analytics to identify potential hotspots, predict disease propagation trends, and generate actionable recommendations for public health stakeholders. Built using a full-stack architecture, the application leverages Flutter for the frontend, Firebase for backend services and cloud database, and the DeepSeek & Gemini API for AI-powered qualitative insights. While existing solutions in countries like India and China have shown the potential of merging AI with geospatial analysis for epidemic tracking and agricultural pest management [1], such integration is limited in Zambia’s public health landscape. This research addresses that gap through a scalable, mobile-centric solution designed specifically for localized deployment. A qualitative evaluation of the prototype indicates strong potential for improving epidemiological surveillance and informing data-driven interventions in Zambia and similar low-income settings.

Cybercrime and social assistance analyzing scam tactics and developing countermeasures case study of developing country zambia

2025-12-25T11:43:06+00:00

This dissertation investigates the increasing prevalence of cybercrime scams targeting social assistance beneficiaries in Zambia, a concern that poses significant risks to vulnerable populations reliant on healthcare and social support systems. The research identifies specific tactics employed by scammers, such as phishing, social engineering, and impersonation, while revealing systemic vulnerabilities exacerbated by inadequate protective measures and a general lack of awareness among citizens. By employing a mixed-methods approach, including qualitative and quantitative data collection through surveys, interviews, and case analyses, the study uncovers critical insights into the demographics most affected by these scams and the psychological impacts they endure. Findings indicate that a considerable proportion of beneficiaries remain uninformed about existing protections, leading to increased susceptibility to fraud. The implications of this research are profound, as it sheds light on the urgent need for targeted educational initiatives and the enhancement of cyber security protocols within social assistance frameworks. The study not only contributes to the existing literature on cybercrime and social welfare but also suggests a model for bolstering the resilience of healthcare systems in developing countries. By addressing these vulnerabilities and implementing robust countermeasures, policymakers can mitigate the impact of cybercrime, ultimately protecting the health and well-being of at-risk populations

Implementation of an IoT Prototype for Real-Time Flood Monitoring and Response Coordination

2025-12-25T14:38:33+00:00

Floods remain one of the most catastrophic natural hazards globally, with disproportionate impacts in under-resourced and infrastructural weak regions. This paper presents the design, development, and evaluation of a low-cost, Internet of Things (IoT)-enabled prototype for real-time flood monitoring and coordinated response, tailored specifically for deployment in rural Zambia. The system architecture integrates the HC-SR04 ultrasonic sensor and ESP32 microcontroller to monitor rising water levels with centimeter-level accuracy. A custom backend, built using Node.js, processes the sensor data and triggers event-based notifications through email alerts. Simultaneously, data are relayed to a lightweight, local web dashboard for visualization and audit purposes. Emphasis was placed on affordability, modularity, offline tolerance, and low power consumption to meet the unique constraints of the target deployment context. Controlled testing scenarios demonstrated reliable performance, with average latency for alerts remaining below six seconds and a sensor accuracy variance within ±1.5 cm. Usability assessments based on the System Usability Scale (SUS) yielded a score of 88.2, affirming the system’s accessibility to non-expert users. The proposed solution holds significant promise as an early warning system (EWS) component in flood-prone, low-resource regions, with potential for integration into broader community-based disaster risk reduction strategies.

Using XAI and Visualization for Clinical Decision-Making: Targeting Mental Health Assessment Using Social Media Data

2025-12-25T16:28:35+00:00

Depression is a major global mental health challenge, often going underdiagnosed due to stigma, delayed recognition, and limited monitoring, reducing opportunities for timely intervention. This study applied Explainable Artificial Intelligence (XAI) and visualization techniques to detect signs of depression from social media data, aiming to improve both predictive accuracy and interpretability. A dataset of 27,977 anonymized public posts was collected from X (formerly Twitter) between 2023 and 2024, cleaned and preprocessed to yield 26,925 posts. Two models were developed: a TF-IDF + Random Forest baseline and a fine-tuned DistilBERT transformer model, trained and evaluated using an 80/10/10 train–validation–test split. DistilBERT outperformed the baseline, achieving 92% accuracy, 89% precision, 91% recall, and a 0.94 ROC-AUC score. Error analysis revealed that false positives often involved sarcasm, while false negatives reflected subtle or metaphorical distress. To enhance interpretability, SHAP (Shapley Additive Explanations) was used for global feature importance, while LIME (Local Interpretable Model-Agnostic Explanations) provided local, case-level insights. Validation by mental health professionals confirmed that 86% of explanations aligned with clinical indicators. Visualization tools, including SHAP plots and saliency maps, further improved accessibility of results. Ethical safeguards such as data anonymization and compliance with platform policies were enforced. This work demonstrates that combining transformer-based models with XAI can produce accurate, interpretable frameworks that bridge the gap between AI prediction and clinical reasoning, supporting responsible and trustworthy mental health assessment.

Automatic creation of Wikipedia articles about Zambia utilizing Retrieval-Augmented Generation techniques and fact-based vector databases

2025-12-25T11:05:46+00:00

This study investigates the automatic creation of Wikipedia articles about Zambia through the use of Retrieval-Augmented Generation (RAG) techniques integrated with fact-based vector databases. While Wikipedia serves as a vital open-access knowledge platform, its coverage of Zambia remains inadequate, with many topics underrepresented or missing. Generative AI, particularly Large Language Models (LLMs), presents opportunities for addressing these gaps but is hindered by issues such as factual hallucination and reliance on low-quality, machine-translated web data. To address these challenges, this research proposes a RAG-based approach that grounds content generation in curated, reliable datasets to improve accuracy, contextual relevance, and editorial usability. The study employs a mixed-methods design involving controlled experiments with Zambian university students, implementation of a RAG prototype system, and evaluation of editor acceptance of AI-generated drafts. Key objectives include assessing whether factual resources increase willingness to contribute, evaluating the effectiveness of RAG in producing reliable Wikipedia content, and exploring editor perceptions of AI assistance. By combining technical development with empirical evaluation, this research contributes to both the advancement of trustworthy AI content generation and the promotion of equitable digital knowledge representation for underrepresented regions such as Zambia.

Fostering Agricultural Transformation in Zambia: Implementing Artificial Intelligence Solutions for Sustainable Farming

2025-12-25T14:09:19+00:00

This paper explores the transformative potential of Artificial Intelligence (AI) in addressing persistent challenges within Zambia's agricultural sector, which employs approximately 54% of the population and contributes significantly to the national economy. Despite this importance, agriculture remains hindered by low productivity, erratic rainfall, soil degradation, and limited adoption of modern technologies. AI applications—including crop disease detection, precision farming, weather prediction for climate adaptation, and optimization of water resources—have shown significant promise in improving agricultural productivity and resilience. This paper adopts a mixed-methods approach that integrates field observations, interviews with smallholder farmers, extension officers, and quantitative analysis of secondary data. Findings indicate that AI adoption can boost yields, reduce losses, and support climate resilience when adapted to local contexts. However, barriers such as limited rural connectivity, digital illiteracy, and cost of deployment remain. The paper recommends stronger partnerships between government, universities, and the private sector to foster localized AI solutions. The study concludes that AI can be a key enabler in Zambia’s agricultural transformation when embedded within inclusive, farmer-centered frameworks.

Risk Prediction Through Deep Learning Classifiers for a Child Health Protection Decision Support System

2025-12-25T15:59:26+00:00

This research presents the development and implementation of a Machine Learning Decision Support System (ML-DSS) aimed at enhancing child health protection in Zambia. The system utilizes a predictive framework based on a star-schema database architecture, which includes a fact table containing child-level data linked to various health and educational indicators. Specifically, the ML-DSS focuses on binary classification tasks to assess school dropout risks and stunting risks among children, employing deep learning techniques facilitated by TensorFlow. Key results highlight the model’s performance metrics, demonstrating its potential to inform early interventions in child health and education. The research identifies critical factors influencing dropout rates and stunting, emphasizing the significance of nutrition and school attendance. Despite limitations, including the absence of detailed household financial data, the model provides a robust tool for NGOs to enhance their programming and improve child health outcomes.

AI-Powered System for Simplifying and Analyzing Terms and Conditions

2025-12-25T10:36:11+00:00

Terms and Conditions (Ts&Cs) are foundational legal documents governing digital interactions between users and service providers. Despite containing critical information related to user rights, data privacy, and liability, these documents remain largely inaccessible due to their legal complexity and verbosity. This paper presents an AI-driven mobile application that leverages advanced Natural Language Processing (NLP) and Large Language Models (LLMs), particularly GPT-based architectures, to automate the simplification and risk analysis of Ts&Cs. The system provides end-users with concise summaries, risk flags, and contextual indicators for informed consent. The mobile application supports multiple input modalities, including document uploads, web URL parsing, and app-based term extraction. Evaluation was conducted using ROUGE and BERTScore metrics, achieving high fidelity in semantic summarization. Usability testing demonstrated that the system improves comprehension, fosters transparency, and reduces the time required for users to interpret legal documents. This work contributes to the broader discourse on algorithmic transparency, digital fairness, and ethical AI deployment in consumer protection. Experimental results indicate significant potential for scaling such tools across diverse jurisdictions and languages.

Evaluating Security Standards and Frameworks for IoT-Enabled Smart Environments

2025-12-25T13:34:51+00:00

The rapid expansion of IoT-enabled smart environments, such as smart homes and cities, brings notable benefits in efficiency, convenience, and sustainability. However, these advancements also introduce significant security risks as the growing interconnectivity of IoT devices increases their vulnerability to threats like data breaches, device hijacking, and DDoS attacks. Ensuring the security of these environments is crucial to mitigate risks and implement effective controls. Despite the urgent need for comprehensive security frameworks, a significant gap remains in identifying standards and methodologies that address the unique and evolving security challenges of IoT-based systems. This paper aims to address this gap by conducting an extensive review of existing security standards and assessment frameworks, with a particular focus on NIST's (National Institute of Standards and Technology) special publications on security techniques, including those still under development. By analysing their strengths, weaknesses, and areas of focus, the study identifies which frameworks are most suited for IoT-based smart environments. Additionally, it evaluates the practical application of these frameworks in real-world scenarios, examining their ability to uncover vulnerabilities, assess security postures, and guide the implementation of effective countermeasures. The findings highlight that while traditional security frameworks may not fully address the unique challenges of IoT environments, they can be adapted to meet these needs. This paper provides insights for researchers, industry practitioners, and policymakers and paves the way for future research to develop tailored security standards and frameworks. It also discusses the key challenges facing IoT security and offers a roadmap for advancing the safe, secure, and sustainable deployment of IoT technologies.

Leveraging Information Systems for Climate Change Mitigation and Adaptation: A Case Study of Developing Countries Learning from Developed Countries

2025-12-25T15:42:30+00:00

This dissertation digs into how digital systems can be put to work to tackle climate change challenges in developing regions while taking a leaf out of the book of wealthier nations

AI-Enabled Drought Prediction System for Zambia

2025-12-25T10:02:41+00:00

Drought poses significant challenges to agriculture, water resources, and socio-economic stability, particularly in Zambia. This paper presents an AI-driven drought prediction system that integrates historical climate data, IoT-based real-time inputs, and machine learning algorithms, including ensemble models such as Random Forest, XGBoost, and LSTMs. The system provides accurate and localized forecasts, enabling proactive decision-making by farmers, policymakers, and disaster management agencies. Unlike traditional systems, it emphasizes regional adaptability and dynamic model retraining to ensure reliability under evolving climate patterns. Results demonstrate prediction accuracies above 90%, with ensemble approaches outperforming single models. This research highlights the potential of AI in mitigating the impact of climate change by enhancing resilience in drought-prone regions.

Enhancing Healthcare Delivery through Integrated Data Analytics: Insights from Practical Applications in Zambia

2025-12-25T13:13:53+00:00

The integration of data analytics into healthcare systems has the potential to revolutionize the delivery of health services, particularly in resource-limited settings. This paper explores the impact of data integration and analytics for the Zambian health sector, drawing on practical experiences from working at Zenysis Technologies and supporting the Ministry of Health (MoH). By analyzing challenges, solutions, and outcomes of implementing an integrated data platform, the paper highlights the transformative power of data-driven decision-making in improving healthcare outcomes.

Leveraging Artificial Intelligence for the Enhanced Management of Telecommunications Infrastructure Performance

2025-12-25T15:23:43+00:00

The rapid advancements in mobile service technologies have significantly increased the number of telecommunications infrastructure elements. Despite this growth, infrastructure management largely remains a manual process, with human intervention required to assess the status of various network components. Key performance indicators such as availability, reliability, accessibility, integrity, and traffic management continue to rely on human analysis, often leading to inaccurate or delayed insights. This paper explores the potential for adopting Artificial Intelligence (AI) to enhance the management of telecommunications infrastructure, focusing on how Mobile Network Operators (MNOs) can integrate AI into their operational frameworks. The study particularly examines how AI can be effectively applied in multi-vendor environments, where MNOs typically manage infrastructure from diverse suppliers. By leveraging AI, MNOs can automate the analysis and monitoring of network elements, allowing for real-time, data-driven decision-making. This transformation would not only improve operational efficiency but also drive increased revenue, enhance user experience, expand customer bases, and enable proactive network monitoring. The findings highlight the immense potential of AI in revolutionizing the way telecommunications infrastructure is managed, ultimately ensuring more reliable and optimized network performance.

A Novel Approach Based on Convolutional Neural Networks for Maize Disease Detection

2025-12-25T09:31:00+00:00

This study presents an improved method for detecting and classifying diseases in corn leaves using an enhanced ResNet-18 Convolutional Neural Network (CNN) model. To address the challenge of limited data and enhance the model’s adaptability to real-world conditions, data augmentation techniques that simulate field environments are applied. The model is trained and tested on a curated dataset of maize leaf images, and its performance is evaluated using key metrics such as accuracy, precision, recall, and F1-score. The proposed approach achieves an excellent overall accuracy of 99.16%, significantly outperforming traditional CNN models. This demonstrates its strong performance and reliability, offering a scalable solution for automated plant disease detection, particularly in resource-constrained agricultural settings.

Development of IoT Household Appliances Asset Tracking System

2025-12-25T12:47:06+00:00

This research presents an IoT-based asset tracking system for household electronics, addressing the high theft rates and low recovery prospects of valuable appliances by leveraging affordable IoT devices with GPS technology for real-time monitoring, theft detection, and last-known-location reporting. The prototype, built using a Java Spring Boot server, supports multiple GPS protocols (GT06, Teltonika, Astra Telematics) with robust packet parsing and geolocation decoding, validated through mixed-methods qualitative stakeholder analysis and quantitative field testing, which demonstrated high accuracy and reliability despite minor data integrity issues. The study confirms the feasibility and effectiveness of extending IoT tracking to households, proposing a tiered deployment strategy to match tracker capabilities to asset value for cost-efficiency, with future work focusing on energy optimization and smart home integration.

Integrating Model Agnostic Explainability into Supervised Learning for Credit Scoring using SHAP and LIME

2025-12-25T15:06:50+00:00

Advanced machine learning models offer superior accuracy in credit scoring, but their "black box" nature hinders regulatory compliance and erodes trust. This paper addresses this challenge by presenting a hybrid framework, developed using a Design Science Research (DSR) methodology, to integrate model-agnostic Explainable AI (XAI) into the credit scoring pipeline. The framework applies leading XAI techniques, specifically SHAP and LIME, to a range of supervised learning models. A functional, interactive prototype was developed and tested using credit data from the Zambian market. Experimental results revealed a stark "Accuracy Paradox": models with the highest accuracy (84.6%) achieved a perfect specificity of 1.000 by never predicting the minority class, resulting in an F1-Score of only 0.458 and an ROC AUC worse than a random guess (as low as 0.432). XAI techniques proved crucial for diagnosing these failures and providing clear, feature-based explanations for individual loan decisions. This research contributes a practical, integrated artifact that systematically compares multiple models and explanation methods, bridging the gap between complex ML implementation and the pressing need for fair, transparent, and accountable financial decision-making.

A Hybrid Genetic Algorithm–Tabu Search Approach for AI-Driven Exam Timetabling in Higher Education

2025-12-25T09:02:08+00:00

The increasing scale and complexity of higher education institutions have rendered manual and rule-based examination scheduling methods inadequate for managing resource constraints, student diversity, and institutional policies. This paper presents a hybrid Artificial Intelligence (AI)–driven examination timetabling system that integrates Genetic Algorithms (GA) and Tabu Search (TS) to generate conflict-free, efficient, and equitable timetables. The GA component performs global exploration to identify optimal scheduling permutations, while TS executes local refinement to prevent convergence toward suboptimal solutions. Developed using Django (backend), React (frontend), and MySQL (database), the system automates examination scheduling, optimizes venue and invigilator allocation, and adapts dynamically to academic changes. Empirical evaluation using real datasets from Mulungushi University demonstrated a 97.5% reduction in exam conflicts, substantial improvement in scheduling efficiency, and enhanced satisfaction among students and administrators. The proposed hybrid GA–TS framework provides a scalable and modular foundation for future AI-based scheduling research, contributing to the advancement of intelligent academic management systems in higher education.

Design and Implementation of a Dual-Mode IoT-Based Gas Leakage Detection and Alert System for Residential Safety in Resource-Constrained Environments

2025-12-25T11:51:56+00:00

The increasing frequency of gas-related accidents in residential, industrial, and institutional settings underscores the urgent need for reliable, real-time gas detection and alert systems. This study presents the design and implementation of an intelligent Gas Leakage Detection and Alert System that integrates Internet of Things (IoT) and embedded sensor technologies for continuous environmental monitoring. The proposed system detects the presence of hazardous gases such as liquefied petroleum gas (LPG), methane, and carbon monoxide using calibrated MQ-series sensors interfaced with a microcontroller unit. A Wi-Fi-enabled module transmits real-time data to a remote cloud server, enabling automated notifications via mobile and web platforms to alert users and emergency responders. The prototype includes both audible and visual alerts, with system thresholds optimized for sensitivity and minimal false positives. Experimental evaluation under controlled and field conditions demonstrated high accuracy in detecting gas concentrations above 200 ppm within less than 5 seconds of exposure, ensuring prompt response to potential hazards.
The proposed framework offers a cost-effective and scalable solution suitable for smart home integration, industrial safety, and public infrastructure management. Future improvements will focus on adaptive threshold algorithms, cloud-based analytics, and integration with smart city platforms to enhance predictive safety intelligence

Improving Indexing Visibility for University of Zambia Hosted Journals: Practical Approaches and Workflow Enhancements through Open Journal System Integration

2025-12-25T14:48:43+00:00

Enhancing the indexing readiness of scholarly journals is essential for increasing global visibility, accessibility, and academic credibility. This study improves the discoverability of Diamond Open Access journals at the University of Zambia through targeted technical and editorial interventions. Using a data mining approach, metadata was extracted from Google Scholar, Crossref, and the University of Zambia journal platform (via OAI-PMH) to compare indexing coverage. The work implements ORCID integration, metadata quality enhancements, and lightweight Open Journal Systems (OJS) plugins for automated metadata verification, DOI deposition, and DSpace package export. Initial tests show these tools help identify and correct indexing barriers, especially incomplete metadata and formatting inconsistencies. Findings indicate that technical upgrades and standardized editorial workflows can significantly improve readiness for inclusion in indexing services such as the Directory of Open Access Journals (DOAJ) and African Journals Online (AJOL). This approach offers a scalable model for strengthening institutional journal support and sustaining visibility in resource-limited environments.

Utilizing Artificial Intelligence to Enhance Personalized Learning at Zambian Universities: A Case Study of NIPA

2025-12-25T16:33:45+00:00

The integration of Artificial Intelligence (AI) in personalized learning offers significant potential to transform education in Zambian universities. Given the growing diversity of student populations and the challenges in traditional educational delivery, AI-driven adaptive learning systems present an innovative approach to customizing learning experiences to meet individual student needs. This study investigates the role of machine learning algorithms in analyzing student performance data to personalize course content, suggest relevant resources, and provide real-time feedback, thereby enabling individualized learning paths. Focusing on the National Institute of Public Administration (NIPA) as a case study, the research explores the impact of AI on student engagement, academic achievement, and retention in the context of Zambian higher education. The study also identifies challenges such as infrastructure limitations, concerns around data privacy, and the necessity for faculty training. The findings aim to offer valuable insights into the viability of AI-powered personalized learning systems in resource-limited environments, with practical recommendations for effectively integrating AI into university teaching practices to improve educational outcomes for both students and faculty.

A Cybersecurity Framework for Optimizing Broadband QoS in IoT Systems Using Machine Learning

2025-12-25T08:20:55+00:00

The integration of Internet of Things (IoT) technologies in healthcare, particularly in Intensive Care Units (ICUs), holds transformative potential for patient monitoring and clinical decision-making. However, performance is often limited by high network latency and cybersecurity vulnerabilities, which are especially critical in time-sensitive applications such as remote monitoring and telemedicine. Achieving both ultra-low latency and strong data confidentiality in resource-constrained ICU environments remains a major challenge, as traditional methods fall short of meeting these dual requirements. This paper proposes a machine learning (ML)-driven cybersecurity framework that optimizes broadband Quality of Service (QoS) while ensuring robust data security in ICU-based IoT networks. The framework integrates supervised and unsupervised learning models for dynamic, context-aware adaptation to network conditions and emerging threats. Key features include intelligent traffic prioritization, secure communication protocols, and adaptive bandwidth allocation. Expected outcomes are reduced latency, improved confidentiality, and enhanced reliability of ICU systems. Beyond technical contributions, the framework promotes trust in digital healthcare and advances interdisciplinary research across ML, network optimization, and medical cybersecurity.

Creating Intelligent and Adaptive Systems for Energy – Efficient Smart Home Appliances Using Tiny Machine Learning

2025-12-25T11:16:02+00:00

The rapid proliferation of smart home appliances has intensified global energy demands, necessitating innovative solutions that balance intelligence with sustainability. This research proposes a novel framework for energy – efficient smart home systems using Tiny Machine Learning (TinyML) to enable real – time, adaptive, and privacy – preserving intelligence on ultra – low – power embedded devices. While existing approaches rely on cloud – dependent AI introducing latency, privacy risks, and high energy costs this work advances on – device TinyML to create self – optimizing appliances that dynamically adjust their behavior based on user patterns, environmental conditions, and energy constraints. The study addresses three critical gaps in current systems namely, Static model architectures that cannot adapt to real – world variability, Energy – inefficient deployments due to lack of hardware – aware optimizations and Absence of collaborative learning in microcontroller-scale devices. The methodology integrates, context-aware neural networks that autonomously switch between optimized sub – models (1-bit to 8-bit quantization) using reinforcement learning, energy – bounded execution policies leveraging dynamic voltage / frequency scaling (DVFS) and intermittent computing for energy – harvesting scenarios and a lightweight federated learning framework enabling privacy-preserving knowledge sharing across appliances without raw data exposure. This research contributes to sustainable computing by redefining how smart homes leverage embedded AI, with broader implications for IoT, Industry 4.0, and green technology. The proposed framework will be released as open – source tools to accelerate TinyML adoption, alongside patent-pending techniques for adaptive edge intelligence.

Generative AI for Bridging the Digital Divide in SADC Higher Education: A Quantitative Study on Student Perspectives.

2025-12-25T14:16:46+00:00

Generative Artificial Intelligence (GenAI) has rapidly transformed higher education by enabling content creation, personalized learning, and academic support. While adoption has advanced in developed contexts, its use in emerging regions remains underexamined. This study investigated student engagement with GenAI specifically ChatGPT, Gemini, and Claude across higher education institutions in the Southern African Development Community (SADC). Using Activity Theory as the analytical lens, a quantitative survey was conducted with 908 students from diverse academic levels and disciplines. The findings revealed high awareness, with 73% of respondents actively employing GenAI for academic writing, conceptual clarification, and the generation of study materials. Despite positive perceptions of usefulness, significant gaps were identified, including insufficient ethical guidance, digital literacy challenges, inconsistent institutional policies, and faculty resistance. These shortcomings hindered responsible and equitable integration. The study underscores the necessity of context-sensitive strategies to support ethical, inclusive, and pedagogically sound adoption of GenAI in higher education. The results provide practical insights for policymakers, institutions, and educators seeking to harness AI responsibly within resource-constrained environments.

Smart Contract-Driven FX Compliance: A Hyperledger Fabric Framework for Real-Time Monetary Policy Enforcement in Zambia

2025-12-25T16:13:19+00:00

Inefficiencies in Zambia’s foreign exchange (FX) ecosystem—marked by fragmented oversight, volatility, and weak enforcement—limit the effectiveness of monetary policy. This paper proposes a blockchain-based framework using Hyperledger Fabric to deliver a secure, permissioned, and tamper-resistant FX management system. Unlike existing blockchain compliance trials and Central Bank Digital Currency (CBDC) pilots such as Nigeria’s eNaira or South Africa’s Project Khokhar, the framework is uniquely tailored for real-time, cross-institution FX compliance in developing economies. Developed through the Design Science Research Methodology (DSRM), the prototype integrates smart contracts, peer nodes, certificate authorities, and compliance logic. Smart contracts automatically enforce daily FX transaction limits, flagging or rejecting violations based on client identity. The architecture employs chaincode for programmable enforcement, an immutable ledger for auditability, and the Raft protocol for efficient ordering. The study is guided by three questions: (i) how smart contracts can enforce real-time compliance; (ii) what performance thresholds for latency, throughput, and compliance accuracy are achievable; and (iii) what scalability challenges emerge for national deployment. A Docker-based testbed achieved block confirmation times under two seconds, compliance accuracy above 97%, and throughput of up to 50 transactions per second beyond technical feasibility, the study highlights adoption barriers—regulatory buy-in, governance, integration costs, and infrastructural constraints— while affirming blockchain’s potential to enhance transparency, stabilize currency markets, and strengthen monetary policy in Zambia and Sub-Saharan Africa.

Assessing Cybersecurity Risks in E-Commerce: Strategies for Threat Mitigation and Protection

2025-12-25T10:47:45+00:00

The exponential growth of e-commerce has also come with a multitude of cybersecurity

problems, subjecting business and consumers to huge vulnerabilities of data breaches, phishing, and payment fraud. Despite advances in technology in security, the majority of the e-commerce sector is vulnerable to new types of cyber threats, inadequate regulatory systems, and poor risk management systems. This paper provides an exhaustive overview of the evolving cyber defense landscape in online shopping, with a focus on the most prevalent threats and appropriate mitigation strategies. It identifies crucial research shortcomings in the areas of emerging threats, the implementation of AI-based security technologies, and end-user awareness as an area to enhance security levels. Through the recognition and completion of these gaps, this study presents a comprehensive framework that will promote e-commerce security measures and online payment systems' trust. The findings have significant implications for policymakers, business leaders, and security professionals, presenting actionable information to improve e-commerce security and establish consumer confidence in online purchases.

Exploring the Impact of Cloud Computing on Business Process Management: Opportunities and Challenges

2025-12-25T13:57:55+00:00

Cloud computing has revolutionized Business Process Management (BPM) by offering businesses cost-effective, flexible, and elastic solutions. Cloud offerings enhance organizations' quest for operational excellence through the ability of rapid workflows, productivity, and responsiveness. Issues relating to information security, compliance with regulatory requirements, and interoperability with in-place systems have remained a hindrance to the adoption of cloud technologies globally. The study discusses cloud computing's impact on BPM—its implication on process automation, collaboration, and decision-making. The paper bridges the literature gaps for cloud BPM adoption process understanding, the ensuing security threats, and the issue of regulation. Conjoining opportunity and challenge analysis, the work proposes an integrative framework for organizations to leverage the use of cloud computing to build better processes

Optimizing Taxi Services through AI-Driven Customer Behavior Modeling and Geo-fenced Hotspot Recommendations for Drivers

2025-12-25T15:47:56+00:00

The shift in consumer behavior in Zambia, from traditional in-store shopping to the rapidly growing demand for delivery and ride-hailing taxi services, has created an urgent need for more efficient and time-saving transportation solutions. Urban centers such as Lusaka, Ndola, and Kitwe have experienced significant increases in ride-hailing activity, driven by the convenience and accessibility of digital platforms. While these services improve customer access and transparency in pricing, they have also introduced new operational challenges for drivers. Longer wait times for customers, increased travel distances to reach riders, and frequent cancellations have led to rising operational costs, particularly in fuel expenditure. Drivers often absorb these costs without guaranteed revenue, reducing profitability and discouraging long-term sustainability in the taxi industry. This research investigates the application of Artificial Intelligence (AI) to address these inefficiencies through a customer request behavioral model that identifies demand patterns, creates geo-fenced hotspots, and provides drivers with intelligent recommendations on optimal positioning. By leveraging Zambia-specific datasets and simulating ride request trends, cancellation behavior, and fuel costs, the study demonstrates how AI can reduce idle driving, improve service efficiency, and enhance driver profitability. Statistical tools, including line graphs, histograms, scatter plots, and pie charts, are applied to consolidate findings and provide empirical evidence of the model’s effectiveness. The results indicate that AI-driven geo-fencing has the potential to significantly minimize operational costs, increase customer satisfaction through shorter waiting times, and create a scalable framework for optimizing both taxi and delivery services in Zambia’s evolving transport sector.

AI-POWERED INCIDENT RESPONSE MODEL FOR DIGITAL FINANCE IN ZAMBIA

2025-12-25T10:14:07+00:00

Zambia’s digital finance ecosystem including mobile money, fintech platforms, and online payments has expanded rapidly, raising exposure to fraud, insider compromise, and data breaches. Current incident response remains reactive and human dependent, leading to delayed containment. This study relies exclusively on secondary data; peer-reviewed research, industry reports, and policy frameworks to propose a contextualized, AI-powered incident response model for Zambia’s financial sector. The framework integrates machine learning into existing Security Information and Event Management (SIEM) workflows, emphasizing modularity and phased adoption. Synthesized evidence highlights gains in detection accuracy and response speed, while also identifying challenges in data availability, organizational trust, and regulatory clarity. The main contribution is a secondary data-driven conceptual architecture designed for resource-constrained contexts, providing a foundation for pilot evaluations and regulatory engagement. This work contributes to building resilient digital finance systems in Sub-Saharan Africa.

Enhancing Short-Interval Controls in Open-Pit Mining through the Integration of Business Intelligence Analytics and the Internet of Things: A Case Study of Lumwana Mining Company

2025-12-25T13:20:11+00:00

The integration of Business Intelligence (BI) analytics with Internet of Things (IoT) technologies offers significant potential to optimize operational efficiency in open-pit mining operations. This paper investigates the application of these technologies to enhance Short-Interval Controls (SIC) at Lumwana Mining Company, one of Africa's largest open-pit mining sites. The research focuses on leveraging IoT-driven sensor networks, machine telematics, automated data acquisition systems, and BI analytics to derive actionable insights. By employing advanced AI-driven analytics, including predictive modeling and data visualization, the study aims to optimize production scheduling, improve equipment productivity, and reduce idle time. A mixed-methods approach is adopted, combining quantitative analysis of real-time operational data with qualitative insights from industry experts to evaluate the effectiveness of BI-IoT integration for SIC. The findings will contribute new knowledge to the field of data-driven mining operations, offering a flexible and scalable model for improving efficiency, sustainability, and cost management in open-pit mining. Practical recommendations derived from this study will assist mining engineers, data analysts, and decision-makers in effectively implementing digital transformation strategies to optimize mining processes.

Leveraging Big Data Analytics for Predictive Modeling and Forecasting in Agriculture in Zambia

2025-12-25T15:29:28+00:00

Big Data is defined using characteristics and concepts beyond size, pinpointing to the volume, velocity, variety, and veracity of the data. The integration of big data analytics in agriculture is revolutionizing farming practices, crop management, and decision-making processes. Much of the existing research has utilized limited datasets and simplistic analytical methods, such as basic statistical approaches and opaque machine learning models, which hinder clear interpretation by farmers and stakeholders. The study aimed to develop a predictive model and forecasting accuracy using data analytics that will improve crop yield in Agriculture, applied advanced data analytics approaches with tree-based machine learning techniques to pinpoint key factors that influence agricultural productivity and used key factors to build a model that predicts crop yield. The study implemented experimental methodology. Utilizing the LightGBM framework - a gradient boosting model known for its interpretability, analyzed an amalgamation of data from surveys, farm records, and climatic information to assess feature importance. It also integrated diverse datasets from governmental reports and agricultural archives. This analysis included various socio-economic factors such as access to water, soil quality, type of seeds, weather pattern, educational levels of farmers, and market access, which were identified as critical variables affecting agricultural success. The LightGBM model not only achieved high accuracy and reliability but also provide transparent insights, outperforming other methods like XGBoost, decision trees, and random forests in our evaluations.

A Systematic Literature Review on inhibitors and challenges in Cloud Computing as a Platform for Big Data Storage and Analysis

2025-12-25T09:39:14+00:00

The ability to handle large volumes of data at the same time, allowing a flexible, reliable, and a more secure access to this data, has proved that cloud computing is appropriate even more in Big Data analysis. Effective and efficient organization of large-scale data analysis illustrates a critical challenge. Lately, cloud computing has played a significant role in the storage and processing of large data sets. This systematic literature review (SLR) aims to review the existing research studies on cloud computing as a platform for Big Data analysis. The search was done through IEEExplore, Springer, Science Direct, Arxiv, ACM, Tandfonline, DAOJ, Emerald insight and hindawi databases in order to sprout out any article related either directly or indirectly to cloud computing. This SLR examined the research studies published between 2020 and 2025 within the popular digital libraries. About one hundred and fifty (150) articles were identified but twenty (20) papers were selected after a meticulous screening of published works to answer the proposed research questions. This systematic literature review presents the key issues of cloud computing in big data processing, including elements of platforms, cloud architecture, cloud database and data storage scheme. Finally, it discusses the open issues and challenges, and deeply explores the research directions in the future on big data analysis in cloud computing platforms.

Drone-Based Remote Sensing for Monitoring Illegal Waste Sites: A Scalable Framework for Urban Environmental Management in Zambia

2025-12-25T12:59:12+00:00

Illegal waste dumping presents a persistent challenge in developing nations, posing severe environmental, social, and public health risks. Traditional inspection and monitoring systems remain inefficient, resource-intensive, and difficult to scale in fast-growing urban environments. This paper presents a novel drone-based remote sensing framework designed to detect, classify, and map illegal waste sites in real time using high-resolution imagery and artificial intelligence. The proposed system integrates object detection models with Geographic Information Systems (GIS) for spatial analytics and visualization, supported by a severity threshold algorithm that prioritizes high-risk sites based on image confidence and area metrics. Field deployment in Lusaka demonstrated an overall detection accuracy of 87%, with a mean precision of 0.89 and recall of 0.85, validating the system’s technical feasibility and robustness. Additionally, a public-facing reporting interface and an administrative dashboard were developed to strengthen community participation and enhance enforcement efficiency. The results show that this approach provides a scalable, cost-effective, and data-driven solution for environmental monitoring, with the potential for broad implementation across sub-Saharan Africa and other developing regions.

Intelligent and Secure Data Access for Non-IT Staff: AI-Powered NLP with Advanced Cybersecurity Controls

2025-12-25T15:11:43+00:00

Data is key to every organization in making informed decisions and more often than note, access to data is expected to be timely and with less challenges. Unfortunately, non-IT staff struggle to access data easily and have to rely on IT staff who are often overloaded with redundant data requests. Conventional data access tools such can be limited and sometimes require IT skills to effectively use them. This creates a need to find a solution that can help non-IT staff who are the main users of data to access data easily without the need for sophisticated systems that need too much technical know-how. AI has significantly improved and has shown potential to solve the problem of data access by non-IT staff through natural language process. Much as AI can solve data access problem, it poses security concerns which ought to be addressed before AI can be used for easy data access. This study collected data from both IT and non-IT staff to understand data access challenges, impracticality for IT to timely attend to ad hoc and redundant data requests, openness to AI solutions and any concerns. Data collected reviewed that 64% of 37 non-IT staff had challenges in data access, 80% were open to AI solutions and 68% had security concerns with AI. A prototype was developed to show feasibility of the system and the results from the tests carried out showed that the solution is feasible and able to solve the problems of data access.

A Hybrid Machine Learning Model for TB/HIV Progression Prediction Using Resource-Constrained Electronic Health Record (EHR) Data in Zambia

2025-12-25T09:14:13+00:00

Tuberculosis (TB) remains a leading cause of mortality among people living with HIV (PLHIV) in Zambia, posing a major challenge to an already strained health system. Zambia’s national electronic health record (EHR) systems, contains valuable longitudinal data that could support predictive tools for early TB intervention. However, issues such as data sparsity, limited analytical capacity, and poor interpretability of machine learning (ML) models have slowed clinical adoption. This study proposes a hybrid ML framework that integrates Random Forest (RF), eXtreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM) networks, enhanced with SHapley Additive exPlanations (SHAP) for transparency. The Design Science Research (DSR) methodology guides iterative model development, evaluation, and deployment. Preprocessing employs Multiple Imputation by Chained Equations (MICE) for missing data, Min-Max normalization for scaling, and SMOTE for class balancing. Data mapping from EHRs has been completed, and a preprocessing pipeline is under development. Initial training and validation are being conducted using synthetic EHR datasets, with performance measured by F1 Score and Area Under the Precision-Recall Curve (AUC-PR). Prototype models will be tested in simulated clinical workflows to assess feasibility and responsiveness. The research contributes a novel ensemble-based approach that fuses static and temporal variables with explainable AI, supporting early HIV/TB progression prediction and clinician trust in low-resource settings. Future work will focus on real-world validation, stakeholder feedback, and integration into national digital health systems.

Design and Implementation of Internet of Things (IOT)-Based Fuel Level Monitoring and Management System

2025-12-25T12:07:48+00:00

Fuel level monitoring in Zambia continues to rely on outdated manual methods, such as dipsticks, which are inaccurate, inefficient, and prone to theft and mismanagement. This study presents the design and implementation of an Internet of Things (IoT)-based Fuel Level Monitoring and Management System (FLMMS) aimed at automating real-time fuel tracking and management. The system employs ultrasonic sensors, a fuel level sensor , Arduino Uno, GSM/GPRS modules, and a dynamic web platform to detect and relay fuel levels and abnormal consumption events. Real-time alerts via SMS and web dashboards enhance operational oversight. Testing showed a 98% accuracy rate and system uptime of 96%. The FLMMS offers a scalable, low-cost, and efficient solution adaptable to Zambia’s infrastructure, bridging critical gaps in fuel security, transparency, and data-driven management.

Industrial Safety Sentinel: A Wearable-Based Real-Time System for Automated Hazard Detection and Machinery Shutdown

2025-12-25T14:54:58+00:00

Industrial environments remain inherently hazardous, particularly where human operators must work in close proximity to heavy or automated machinery. This paper presents the Industrial Safety Sentinel, a novel wearable-based safety framework that proactively mitigates workplace accidents by integrating real-time hazard detection with automated machinery shutdown. The system embeds a microcontroller-driven chip into worker attire, enabling continuous wireless communication with surrounding equipment. Upon detecting encroachment into predefined danger zones, the Sentinel autonomously initiates machine shutdown, thereby reducing reliance on human reflexes and minimizing latency in emergency response. Development followed the Spiral Model methodology, incorporating iterative prototyping, structured risk assessment, user feedback integration, and staged performance evaluation. Experimental validation in simulated industrial scenarios demonstrated a mean hazard response latency of 2.3 seconds, proximity detection accuracy of 97.8%, and operational availability of 99.7%, confirming the system’s reliability and robustness. By bridging wearable sensing with cyber-physical automation, the Sentinel advances a scalable, worker-centric safety paradigm adaptable across diverse industrial contexts, contributing to the next generation of proactive occupational safety systems.

A Deep Learning Model for Corn Yield Prediction Using Spatial and Temporal Features

2025-12-25T08:36:19+00:00

Maize is a staple food crop in Zambia, making accurate yield prediction essential for food security and agricultural planning. This study presents an advanced deep learning approach for maize yield prediction that integrates Sentinel-2 satellite imagery with climate data. We develop a scalable and interpretable hybrid CNN-LSTM model to capture both spatial and temporal patterns of crop growth. The CNN component extracts spatial features from Sentinel-2 multispectral images (including vegetation indices such as NDVI and EVI), while the LSTM component learns temporal dynamics from time-series climate variables (rainfall, temperature, humidity). The model is trained and validated using historical yield records from major maize-growing regions in Zambia, demonstrating high predictive accuracy and outperforming traditional yield estimation methods. Accurate yield forecasts from this model enable early warnings of potential crop shortfalls, allowing farmers to take timely action to mitigate losses. Additionally, the predictions provide policymakers with insights for managing grain reserves, market supply, and food security strategies. By leveraging deep learning and remote sensing, this work offers a decision-support tool that contributes to sustainable agricultural practices and climate resilience in SSA, bridging the gap between academic and practical applications.

Cultural Influences on Teachers' Attitudes towards AI Integration: A Case Study of Katoba Secondary School in Chongwe District, Zambia

2025-12-25T11:35:34+00:00

This qualitative case study, guided by the Diffusion of Innovations (DOI) Theory and underpinned by an interpretivist paradigm, aimed to explore the multifaceted cultural influences on teachers' attitudes towards Artificial Intelligence (AI) integration at Katoba Secondary School in Chongwe District, Zambia. With a sample size of 8 participants, the study delved into how their beliefs, including pedagogical views and trust in technology, shaped their initial perceptions of AI as an educational tool. Furthermore, it investigated how teachers' privacy values, influenced by local socio-cultural norms, affected their readiness to adopt and integrate AI technologies into their instructional practices. Finally, the research examined how current professional development methods and school-level cultural practices either facilitated or hindered the successful integration of AI tools into daily classroom routines. The study revealed how teachers' deeply held pedagogical beliefs, privacy values, and the effectiveness of professional development and school-level cultural practices collectively served as either enablers or barriers to the successful integration of AI tools into their daily instructional routines. These crucial insights informed culturally responsive policies, curriculum development, and professional development strategies for fostering effective AI integration in Zambian education and comparable sub-Saharan African contexts.

Geospatial Mapping and Assessment of Weather Stations for Enhanced Weather Event Reporting in Zambia

2025-12-25T14:33:43+00:00

The limited spatial distribution of weather stations in Zambia continues to hinder timely and accurate weather reporting, particularly in remote and rural regions. This inadequacy affects sectors such as agriculture, disaster risk management, and climate adaptation planning. This paper proposes a geospatial assessment framework that leverages buffer analysis to identify spatial gaps in Zambia’s weather monitoring infrastructure. Each weather station was assigned a 100 km coverage radius, and a subtraction algorithm was applied against the national boundary to expose uncovered regions. These gaps were visualized in a Flutter-based mobile application that integrates real-time station metadata and interactive mapping. Results reveal significant regional disparities, particularly in North-Western and Western provinces. This system supports data-driven decisions on new station deployments and contributes to early warning systems, sustainable development, and digital meteorological services. The work builds upon previous digital spatial mapping systems such as those used for locust detection [1], integrating them with meteorological use cases to enhance Zambia’s climate resilience.

Strengthening Scholarly Publishing at the University of Zambia: Custom OJS Plugins for Metadata Accuracy, DOI Management, and Repository Integration

2025-12-25T16:21:56+00:00

The growing demand for accurate, interoperable, and easily discoverable scholarly outputs necessitates enhanced publication workflows in institutional repositories and journal management systems. At the University of Zambia, where scholarly publishing is completely reliant on digital platforms, Open Journal Systems (OJS) has emerged as a central tool for journal hosting. However, limitations in automated metadata compliance, DOI integration, and repository deposition pose persistent challenges. This paper is guided by a quantitative data mining approach and presents the conceptualization, development, and deployment of three interlinked OJS plugins designed to enhance the publication pipeline: (1) an Automated Metadata Verification Plugin that validates journal article metadata against institutional and international standards (2) a DOI Deposition Plugin that streamlines Crossref DOI registration and metadata syncing in real time; and (3) a DSpace Export Plugin that packages OJS-published content into DSpace-ready simple archive bundles(e.g ZIP) for seamless repository ingestion. The proposed system is to be implemented and tested within selected UNZA-hosted journals especially under the JABS editorial workflow. Preliminary results from an empirical analysis show substantial improvements in metadata completeness, cross-platform visibility, and compliance with repository requirements. Additionally, the plugins promote FAIR (Findable, Accessible, Interoperable, Reusable) principles by ensuring structured and verified data flows across platforms. By automating routine tasks, reducing human error, and bridging siloed scholarly infrastructure, this plugin suite represents a scalable model for institutions seeking to align with international open access standards. The paper concludes with insights on customization for multilingual metadata, and roadmap plans for integration with institutional repositories

Automated Microbial Classification from +Microscopy Images Using Convolutional Neural Networks

2025-12-25T10:59:03+00:00

The rapid identification of microbial species is essential for advancing clinical diagnostics, environmental monitoring, and food safety assurance. Traditional microbial identification methods, though effective, remain constrained by their reliance on manual labor, extended processing times, and dependence on expert interpretation. This study presents the design and implementation of an artificial intelligence (AI)-powered system for microbial identification using microscopic images. The system integrates convolutional neural networks (CNNs) with transfer learning to enhance classification accuracy and efficiency. A diverse dataset of labeled microscopic images was collected and preprocessed using advanced image enhancement and segmentation techniques to ensure data quality. The trained CNN model demonstrated high performance in classifying bacterial and fungal species, with significant improvements in both speed and reliability compared to conventional methods. The system includes a user-friendly mobile interface that allows image uploads, automated classification, and real-time feedback. Moreover, a continuous learning module facilitates dataset expansion through user-contributed images, supporting model evolution and scalability. The proposed framework underscores the transformative potential of AI in microbiology by automating diagnostic workflows, mitigating human error, and expanding accessibility to microbial analysis tools. Ethical considerations regarding data privacy, transparency, and algorithmic bias were also addressed to ensure responsible AI integration in clinical and research environments. Overall, the project demonstrates how AI-driven image analysis can advance microbial identification, contributing to more efficient, accurate, and accessible diagnostic practices

Factors Influencing User Adoption of Artificial Intelligence in the Telecommunications Industry: Challenges and Strategic Approaches

2025-12-25T14:02:34+00:00

The potential of Artificial Intelligence (AI) to revolutionize the telecommunications sector is undeniable, offering opportunities to enhance service delivery, boost operational efficiency and elevate customer experiences. Despite its growing prominence across industries, the widespread adoption of AI within telecommunications remains a multifaceted challenge. This paper investigates the critical factors influencing users' adoption of AI technologies in this sector, with a particular focus on technological readiness, perceived benefits, ethical considerations, data privacy concerns and socioeconomic barriers. The study highlights how telecommunications companies must address these diverse elements to effectively promote AI adoption among users. Specific strategies discussed include enhancing user awareness of AI's advantages, implementing robust data security protocols and ensuring transparency in AI systems. Furthermore, the paper emphasizes the need for tailored AI solutions that cater to a broad spectrum of user demographics, especially those from lower socioeconomic backgrounds, to foster more inclusive adoption. By providing a comprehensive analysis of these factors, the paper contributes to the existing literature on AI adoption in telecommunications and offers actionable recommendations for industry stakeholders. The findings aim to inform decision-makers in the telecommunications industry on how to overcome adoption barriers, drive innovation and enhance the overall service experience through AI integration.

Real-Time Anomaly-Driven Cyber Resilience: An Adaptive Machine Learning-Based Defense Against False Data Injection Attacks in Smart Grids

2025-12-25T15:52:11+00:00

This work proposes a machine learning-driven adaptive framework for real-time detection and mitigation of FDIAs in critical smart grid infrastructure. The adaptive nature of the model addresses evolving False Data Injection Attacks and provides a more secure and viable method of securing critical smart grid infrastructure from the injection of false data attacks. The fast digital transformation of smart grid infrastructure has created cybersecurity vulnerabilities. Conventional detection models are challenged, and the requirement of a complex solution is required to handle evolving attacks on critical smart grid infrastructure. The technical contributions of this research include continuous update of the model based on the evolving attacks. The model can adapt without retraining from scratch. This model is therefore applicable in future implementations of smart grids, where such models can be adopted by countries who wish to implement smart cities and utility companies in developing countries.

AI-Powered Object Detection in Satellite Imagery for Military Reconnaissance

2025-12-25T10:23:13+00:00

Timely and accurate interpretation of satellite imagery plays a vital role in modern military reconnaissance. This paper proposes VisionAI, a robust, AI-powered object detection system built on the YOLOv8 architecture, optimized for detecting military assets such as tanks, aircraft, trucks, and naval vessels. The system was fine-tuned on a custom remote sensing dataset and deployed using Google Cloud’s T4 GPU infrastructure for real-time inference. The model achieved a mean Average Precision (mAP@0.5) of 0.79 for aircraft and maintained high precision and recall across key object categories. VisionAI demonstrates strong resilience to environmental distortions including cloud occlusion, low lighting, and motion blur. This work builds upon previous efforts in lightweight detection frameworks using MobileNetV2 for pest surveillance in locust management campaigns [1], as well as scalable AI pipelines for real-time monitoring in resource-constrained settings [2]. Furthermore, it aligns with recent advances in satellite surveillance and small object detection using cross-scale and pyramid fusion methods [3], [4]. Challenges related to detecting camouflaged or low-resolution naval targets persist, underscoring the need for hybrid approaches combining multispectral data and transformer-based architectures. Ethical considerations around adversarial manipulation and dual-use of AI in military contexts are also discussed. This research offers a cost-effective, adaptable, and ethically aware solution for defense-oriented remote sensing operations in developing regions.

Evaluating AI Models for Solar Irradiance Prediction: A Systematic Review of Strengths, Limitations, and Future Directions

2025-12-25T13:26:52+00:00

The growing need for accurate solar irradiance prediction to optimize solar energy generation has led to the exploration of Artificial Intelligence (AI) models. This study systematically evaluates the strengths, limitations, and future directions of AI models used in solar irradiance prediction. A total of 80 articles, sourced from databases such as Scopus, IEEE Xplore, and Google Scholar, were included based on relevance to key search terms including “solar irradiance prediction,” “Artificial Neural Networks (ANNs),” “Support Vector Machines (SVM),” “Random Forests (RF),” and “Deep Learning (DL).” The articles were selected using a comprehensive review process that ensured the inclusion of high-quality and relevant studies. The analysis and synthesis of the articles revealed that AI models, particularly ANNs, are widely used due to their ability to model complex, non-linear relationships and provide high prediction accuracy. However, limitations such as overfitting, the need for extensive computational resources, and challenges in data preparation were identified, especially with models like SVM and RF. Hybrid models that combine the strengths of different AI approaches were frequently recommended in the literature. Additionally, future directions for improving solar irradiance prediction included the integration of real-time data, satellite-based information, and the reduction of computational costs. This study highlights the substantial potential of AI in enhancing solar irradiance prediction while also pointing out key challenges. It concludes with recommendations for the development of hybrid models, better computational efficiency, and the use of real-time and satellite data to improve the scalability and accuracy of solar energy forecasting.

Leveraging Blockchain for Cyber Security in Zambia's ICT Sector: A Study on Data Transaction Integrity and Protection

2025-12-25T15:34:43+00:00

This study explores the use of blockchain technology to enhance cybersecurity in Zambia’s ICT sector, focusing on improving data transaction integrity and protection. Given the rising threat of cyberattacks and data breaches, the research investigates how Distributed Ledger Technologies (DLTs) can secure digital transactions by decentralizing data storage and ensuring tamper-resistance. The findings indicate that blockchain outperforms traditional security measures by offering superior transparency, data integrity, and accountability. However, challenges such as limited awareness, regulatory gaps, and technical barriers hinder its adoption. The study concludes that with proper regulatory frameworks, educational campaigns, and pilot projects, blockchain can significantly strengthen Zambia’s cybersecurity framework. Addressing these challenges is crucial for realizing blockchain’s full potential in safeguarding digital transactions and mitigating cyber risks in the ICT infrastructure.

AI Model for Improving Social Protection Delivery in Zambia

2025-12-25T09:49:19+00:00

Social Protection programmes in Zambia aim to reduce the levels of poverty and improve the lives of the vulnerable in the communities. However, the two most common challenges in the Social Protection sector in Zambia are fragmented data and poor targeting of beneficiaries. This report describes the development of an AI model prototype for improving Social Protection delivery in Zambia. AI models such as machine learning are discussed in helping solve the fragmented data problem through integration of data from different management information systems as well as improve the targeting challenge using algorithms. The legal, technological and ethical issues in relation to the prototype being developed are also discussed. Furthermore, the report discussed the future works on the enhancements of the model and its functionality.

Effect of Human Resource Management Practices on Employee Performance in Public Sector Organizations: A Systematic Review Utilising the UTAUT Framework

2025-12-25T13:06:40+00:00

This systematic review investigates the relationship between Human Resource Management (HRM) practices and employee performance within public sector organizations, framed by the Unified Theory of Acceptance and Use of Technology (UTAUT). The primary aim is to understand how specific HRM practices influence employee performance and explore the mediating role of technology adoption and acceptance in this context. A comprehensive literature search was conducted using multiple academic databases, applying targeted keywords related to HRM practices, employee performance, public sector organizations, and UTAUT. The collected studies were systematically analysed to extract relevant findings and insights. The review reveals that effective HRM practices significantly enhance employee performance in public sector organizations. It underscores the critical role of user-friendly technology, as emphasized by UTAUT, in facilitating HRM processes and fostering employee engagement. This study emphasizes the importance of aligning HRM strategies with technological advancements to optimize employee outcomes. However, limitations concerning the exclusion of grey literature and the generalizability of findings to the private sector are acknowledged. Future research is encouraged to further explore these relationships across diverse organizational contexts.

IoT-Based Traffic Management System

2025-12-25T15:18:48+00:00

Rapid urbanization in developing cities has intensified road congestion, increased travel delays, and elevated carbon emissions, underscoring the need for intelligent and adaptive traffic control systems. This study presents the design and evaluation of an Internet of Things (IoT) and Artificial Intelligence (AI) - based traffic management framework developed for urban intersections in Lusaka, Zambia. The proposed system integrates ESP32-based sensor networks for real-time vehicular data acquisition, a cloud-driven processing infrastructure (Firebase), and a locally hosted AI engine employing a Random Forest Regressor for adaptive signal optimization. The system supports both automated and manual traffic control through a responsive Next.js dashboard with a latency below two seconds. Experimental simulations across three intersections revealed a 44% reduction in average vehicle waiting time, a 27% improvement in throughput, and a 27% decrease in estimated fuel consumption. These findings demonstrate the framework’s capacity to enhance mobility efficiency, reduce congestion-related emissions, and promote sustainable urban transport. The paper contributes a scalable, low-cost, and context-aware smart traffic management model adaptable to the infrastructure realities of developing cities. All experiments were conducted in a hardware-in-the-loop simulation environment using benchtop signal heads; no on-road trials with live vehicle traffic were performed.

A Modular AI-Driven Framework for Automating HR Case Processing: OCR, NLP, Sentiment Analysis and Imbalanced Learning

2025-12-25T09:23:45+00:00

This study investigates the feasibility and effectiveness of integrating Optical Character Recognition (OCR), Regular Expressions (RegEx), Aspect-Based Sentiment Analysis (ABSA), and supervised learning to automate structured and semi-structured Human Resource (HR) case processing in African public-sector contexts. The proposed modular pipeline comprises OpenCV-based pre-processing for noise reduction, skew correction, and ROI detection; tuned Tesseract 4.1 OCR; RegEx-driven attribute extraction; ABSA for narrative sentiment; and Balanced Bagging Random Forest classification with SMOTE-ENN applied to enhance minority-class representation and improve sensitivity to rare but decision-critical HR outcomes. Evaluated on anonymised promotion and transfer cases from Zambia’s Teaching Service Commission, the system achieved 95% OCR accuracy, F2-Score of 0.97, weighted F1 = 0.95, PR AUC = 0.98, and minority-class F1 = 0.36, demonstrating improved detection of low-frequency, high-priority HR cases. End-to-end processing reduced manual timelines from ~3 days to <9.4 s per two-page case batch, scaling to 1,312 records in 3.3 s on commodity hardware, enabling timely decision support in resource-constrained environments. The framework’s domain-specific integration and ethical alignment provide a scalable, adaptable solution for HR digitisation in policy-bound, resource-limited sectors, supporting compliance with governance and transparency requirements.

Development of an Intelligent Fire Prevention System Using IoT-Enabled Sensors and Automated Countermeasures

2025-12-25T12:36:22+00:00

Fires pose a significant threat to communities worldwide, with devastating consequences that include the loss of lives, destruction of property, and long-term environmental impact. In Zambia, the rising incidents of fires, often caused by gas leaks, cylinder explosions, electrical faults, and unattended fires, highlight the urgent need for effective fire prevention and early detection systems. Traditional fire safety measures, while useful, often fail to address the complexities and speed with which fires can spread, especially in residential and industrial settings. This paper proposes an innovative, smart fire prevention system leveraging the Internet of Things (IoT) to enhance both fire detection and prevention efforts. The system integrates multiple IoT sensors, including gas leak detectors, smoke detectors, heat sensors, and a fire suppression mechanism, all connected to an Arduino board. Upon detecting a threat, the system immediately alerts occupants, local authorities, and emergency responders through automated notifications, and automatically initiates countermeasures such as activating extinguishing mechanisms. By combining real-time monitoring and automated intervention, the system offers a comprehensive, cost-effective solution adaptable to both residential and commercial applications.

Integrating Augmented Reality Into Zambia’s Education System: A Case Study of an AR-Based Geometry App and Its Impact on Learning Outcomes

2025-12-25T15:00:38+00:00

This research investigates the use of Augmented Reality (AR) to enhance the teaching and learning of geometry in Zambia’s education system. Aligned with national efforts to modernize classrooms and strengthen STEM competencies, the research focuses on developing and evaluating an AR-based application that visualizes three-dimensional geometric shapes in an interactive format. The application allows students to engage with cones, pyramids, prisms, and cuboids through real-time visualization, spatial interaction, and annotated overlays anchored to image markers. Evaluation was conducted in school settings using questionnaires administered to teachers and students to identify instructional challenges, learner preferences, and pedagogical needs. Results showed that students face difficulties in visualizing 3D shapes and understanding spatial relationships, while teachers highlighted the need for interactive tools to improve engagement and conceptual understanding. Both groups expressed a strong preference for hands-on, visually rich learning methods that connect geometry to real-world contexts. The findings demonstrate that integrating AR into classrooms can revitalize traditional teaching practices, enhance comprehension of geometry, and foster deeper learner engagement. This research provides insights for policy-makers, educators, and curriculum designers seeking scalable and context-sensitive innovations for education in developing regions.