http://dissertations.umu.ac.ug/xmlui/handle/123456789/18 Mon, 01 Jun 2026 18:38:42 GMT 2026-06-01T18:38:42Z http://dissertations.umu.ac.ug/xmlui/handle/123456789/1876 A contextual framework for data cleaning in clinical research Kavuma, Pius This study presents the design, development, and validation of a contextual data cleaning framework tailored for clinical research settings in low-resource environments, using the DPSP (Dihydroartemisinin-Piperaquine and Sulfadoxine-Pyrimethamine) trial at Masafu Hospital as a case study. The research was motivated by persistent data quality challenges—such as missing values, inconsistencies, human errors, and tool limitations—that often compromise the validity and reliability of clinical research outcomes. Employing a user-intervention methodology, the study integrated qualitative insights from data managers, clinical teams, and analysts with quantitative assessment techniques to ensure that the proposed framework aligns with real-world practices. The framework was structured into distinct phases, including data profiling, preprocessing, modular cleaning, enhancement, and quality scoring—each mapped to address specific data integrity issues. Validation on the DPSP dataset demonstrated a significant improvement in data accuracy (from 75% to 94%), completeness (from 68% to 90%), and consistency (from 70% to 93%), confirming the framework’s effectiveness and usability. SQL driven automation further improved scalability and reduced human error. The study contributes to the literature by offering a novel, context-sensitive approach that balances domain expertise with technical rigor. It recommends future work to expand the framework’s applicability to unstructured data types and to assess its operational integration and cost-effectiveness. Overall, the framework serves as a practical tool for improving data quality in clinical trials and enhancing the credibility of health research in resource-constrained settings. Julius Muganji Mon, 01 Sep 2025 00:00:00 GMT http://dissertations.umu.ac.ug/xmlui/handle/123456789/1876 2025-09-01T00:00:00Z http://dissertations.umu.ac.ug/xmlui/handle/123456789/1875 Predictive analytics and its impact on tax audit case selection in revenue authorities of developing economies Seremba, Nicholas Revenue mobilization, especially in developing economies, is an arduous task for tax authorities. Uganda Revenue Authority (URA) uses manual, intuitive methods for Corporate Tax audit case selection, which could be time-consuming and ineffective. The study aims to gain insights into the efficiency of predictive analytics in the selection of tax audit cases based on historical tax audit data and taxpayer attributes. Tax officials of the Uganda Revenue Authority (URA) were interviewed, and data analyzed using machine learning algorithms after being trained on 5 years' worth of audit data (2017–2021) provided by URA. The objective was to evaluate and validate the application of predictive analytics for enhanced audit case selection at URA in order to better identify high-risk taxpayers, improve resource allocation and address the operational inefficiencies associated with current audit selection methods. Three models were tested: an initial leaked model (~99% accuracy, AUC 0.99, but unrealistic), a baseline logistic regression model (~62% accuracy, AUC ~0.68), and a final tuned XGBoost model (~75% accuracy, AUC ~0.80, precision ~75%, recall ~85%). The XGBoost model demonstrated substantial improvement over the baseline, correctly flagging a higher proportion of high-risk entities with practical accuracy and balanced performance. The results indicate that high-risk entities selected by the predictive model exhibit a significantly higher likelihood of non-compliance compared to those selected through traditional methods. This demonstrates the potential of predictive analytics to increase audit efficiency and yield while optimizing scarce enforcement resources. This paper joins a recent stream of literature on the application of data mining in tax administration. It contributes to the understanding of how predictive analytics can enhance tax audit efficiency in developing economies, and specifically provides empirical evidence and guidance for policymakers in Uganda and other similar contexts considering data-driven audit selection frameworks. Keywords: Predictive Analytics, Tax Audit, Audit Efficiency, Revenue Mobilization, Uganda Revenue Authority Andrew Lukyamuzi Fri, 01 Aug 2025 00:00:00 GMT http://dissertations.umu.ac.ug/xmlui/handle/123456789/1875 2025-08-01T00:00:00Z http://dissertations.umu.ac.ug/xmlui/handle/123456789/1874 A model for improved adoption of telemedicine in low-resource countries Nassaazi, Sarah The ever-growing advancement in technology has opened up opportunities for healthcare delivery, especially in low-resource environments where access to quality healthcare services is limited. Telemedicine, the remote provision of medical services through telecommunication and information technology, holds incredible potential for bridging the healthcare gap in these underserved areas. However, the adoption of telemedicine in low-resource environments faces unique challenges and requires a tailored approach. This study aimed at designing a Telemedicine Adoption Model for low-resource environments. Furthermore, the study identified factors that affect the adoption and successful implementation of telemedicine in resource-constrained settings. The research employed a mixed-methods approach, combining qualitative interviews and quantitative surveys with healthcare providers, administrators, and patients to gather comprehensive data. The preliminary research findings indicate that several key factors play a pivotal role in telemedicine adoption in low-resource environments. These factors include infrastructure accessibility, technical capability, perceived usefulness and ease of use, social acceptance, regulatory framework, cost-effectiveness, patient trust and satisfaction. The Telemedicine Adoption Model developed through this study incorporates these factors into a comprehensive framework, providing a guide for successful telemedicine implementation in low-resource environments. The proposed Telemedicine Adoption Model can serve as a valuable resource for policymakers, healthcare providers, and organizations seeking to leverage telemedicine to improve healthcare access and results in resource-constrained settings. By considering the special challenges and opportunities of low-resource environments, the TAM offers practical experiences and guidelines for implementing telemedicine solutions that are both effective and sustainable. Additionally, this study contributed to the development body of information on telemedicine adoption and highlighted the significance of context-specific approaches in overcoming barriers to healthcare access and delivery in low-resource environments. This study established a number of critical success factors which included: Attitude towards change (SD=0.731, p-value=0.045), Project planning and management (SD=0.986, p value=0.031), Commitment to Change (SD=0.233, p-value=0.027), Technology- task fit, complexity and training (SD=0.111, p-value=0.019), Management Commitment (SD=0.867, p-value=0.008), Management support (SD=0.568, p-value=0.022), Triability SD=0.981, p value=0.039), Relative advantage (SD=0.998, p-value=0.044), and user satisfaction with the system (SD=1.334, p-value=0.048). x A model for improved adoption of Telemedicine in Low Resource Countries like Uganda was developed. This model has factors such as Organizational Affiliations (ẋ=4.333), Management Commitment and Support (ẋ=3.933), User involvement and triability (ẋ=3.67), telemedicine Policies and guidelines(ẋ=4.5), Technological (ẋ=4.2), Financial(ẋ=4.4) and Human Resources (ẋ=3.67), User acceptance of telemedicine(ẋ=4.133), Organizational structure and culture (ẋ=3.67), Relative Advantage (ẋ=4.6), Hospital Management and staff(ẋ=3.47), IT department of the hospital (ẋ=3.93) and NH Telemedicine Model Outcomes (ẋ=4.8). The developed model for improved adoption of telemedicine in low-resource countries was evaluated by 15 health informatics experts who asserted that the developed model was complete Joseph Brian M. Kasozi Fri, 01 Aug 2025 00:00:00 GMT http://dissertations.umu.ac.ug/xmlui/handle/123456789/1874 2025-08-01T00:00:00Z http://dissertations.umu.ac.ug/xmlui/handle/123456789/1873 Designing a machine learning framework for fraud detection in digital payments Namugera, Francis Xavior Kasozi Joseph Brain Mon, 01 Sep 2025 00:00:00 GMT http://dissertations.umu.ac.ug/xmlui/handle/123456789/1873 2025-09-01T00:00:00Z