UCU Digital Institutional Repository(UCUDIR)
Welcome to the Uganda Christian University Digital Institutional Repositoy (UCUDIR). This is the University's official Institutional Repository. It aims to collect, preserve and showcase the intellectual output of staff and students of UCU. This growing collection of research includes peer-reviewed articles, book chapters, working papers, theses, and more.
- The Repository ingests documents, audio, video, datasets and their corresponding Dublin Core metadata
- The aim is to open up this content to local and global audiences, with have optimized well for Google Scholar so your items here shows up on Google Scholar searches
- we also issue permanent urls and trustworthy identifiers, including optional integrations with handle.net and DataCite DOI
Not Registered? click here to Register or or if already registered: Click To submit your Item ::For more information visit any UCU Library branch
Communities in UCUDIR
Select a community to browse its collections.
- Items in this community present a concise summary of information that can help readers understand, and likely make decisions about, government policies.
- This is a collection of Masters and Doctoral theses and dissertations submitted in electronic format to the Uganda Christian University.
Recent Submissions
Prevalence and factors associated withnon-compliance to iron and folic ACID supplementation among pregnant women receiving antenatal care at Mbale regional referral hospital
(Uganda Christian University, 2026-10-15) Namukona, Sandra
BACKGROUND: In low-income countries, anaemia in pregnancy is a significant cause of maternal morbidity and mortality. The World Health Organisation and the Uganda ministry of health recommend a daily intake of sixty (60) mg of iron and 400 µg of folic acid supplementation for all women during pregnancy and a daily 400 µg of folate a month prior conception. This study aimed at assessing Prevalence and Factors associated with noncompliance toiron folic acid
supplementation among women attending antenatal care at Mbale Regional Referral Hospital. METHODS: This was a cross-sectional study conducted at the Mbale Regional Referral Hospital. Participants were recruited by consecutive sampling and data was collected using a structured questionnaire. The primary outcome was noncompliance to iron folic acid supplementation defined as taking at least less than 70% of the prescribed dose of the Iron Folic Acid Supplementation tablets in the week preceding the interview, an equivalent of less than 5 days/week. Multivariable logistic regression was done to assess for factors associated with noncompliance with Iron Folic Acid Supplementation. Data were analyzed in Stata version 14 software. RESULTS: The study found a high level of non-compliance (37.1%) to iron folic acid Supplementation among pregnant women attending antenatal care at Mbale Regional Referral Hospital. Pregnant women who did not know their HIV status [Adjusted Odds Ratio (AOR): 2.36; 95% CI: 1.39-4.00] were more likely to be noncompliant to iron folic acid Supplementation. While Pregnant women who had Previous birth by C-Section AOR: 0.55; 95% CI: 0.32-0.95]and pregnant women who found the smell of iron folic acid Supplementation being non irritating [AOR: 0.60; 95% CI: 0.39-0.92]. were less likely to be noncompliant to iron folic acid supplementation. CONCLUSION AND RECOMMENDATION: The study found a high level of noncompliance (37.1%) to iron folic acid Supplementation among pregnant women attending antenatal care at Mbale Regional Referral Hospital. The group at risk for noncompliance was pregnant women who did not know their HIV status. I recommend further investigation into the hinderance to noncompliance to IFAS and women not knowing their HIV status.
Multimodal deep learning-based detection of fetal abnormalities from prenatal ultrasound images and clinical data in Ugandan health facilities
(Uganda Christian University, 2026-10-12) Ssemukuye, Timothy
Background: Uganda's high maternal mortality ratio (189 per 100,000 live births in 2022) is linked to inadequate prenatal diagnostic capacity. Shortages of trained sonographers and variable equipment quality limit the utility of ultrasound in resource-constrained district hospitals. Objective: To develop and evaluate a multimodal deep learning system for detecting fetal abnormalities from prenatal ultrasound scans and associated clinical data using the UCSF-IDRC Uganda Prenatal Study dataset. Methods: A late-fusion multimodal architecture (Proposed Multimodal Diagnostic Model) was developed combining an EfficientNet-B0 image branch (ImageNet pretrained, top 30% fine-tuned) with a 15-feature clinical MLP branch. The dataset comprised 2,623 deduplicated ultrasound images (420 normal, 2,203 abnormal) from Masafu General Hospital, Busia. Patient-level stratified splitting produced training (n=1,925), validation (n=347), and test (n=351) sets. Training used AdamW optimisation with cosine annealing over 20 epochs on Apple MPS hardware. Results: On the held-out test set (n=351), the model achieved AUC-ROC=0.7829, AUPRC=0.9601, sensitivity=79.73%, specificity=54.00%, precision=91.25%, F1=0.8511, and accuracy=76.07%. The best checkpoint was at epoch 12 (val AUC=0.8188). Occlusion saliency analysis confirmed the model attends to diagnostically meaningful image regions consistent with expert sonographic assessment. Conclusions: The proposed multimodal diagnostic model demonstrates clinically meaningful fetal abnormality detection performance in a Ugandan prenatal care context, with sensitivity 32.7 percentage points above the 47% unaided clinical sensitivity reported for district hospital sonographers. The results support feasibility of AI-assisted prenatal diagnosis in resource-limited settings, subject to additional normal scan collection and prospective clinical validation.
Fraud detection in microfinance banking using machine learning. A case study of Ugafode microfinanance (MDI)
(Uganda Christian University, 2026-10-16) Sserwadda, Samson
Fraudulent transactions continue to pose a significant challenge to microfinance institutions as digital banking channels expand, exposing institutions to increasingly sophisticated financial crimes. UGAFODE Microfinance (MDI), like many microfinance institutions in Uganda, relies mainly on rule-based and manual fraud monitoring approaches, which are limited in detecting emerging and evolving fraud patterns. This study was therefore conducted to develop and evaluate a machine learning-based fraud detection model for identifying potentially fraudulent transactions at UGAFODE Microfinance. The study used historical transactional data extracted from UGAFODE Microfinance’s core banking system. The dataset comprised transaction records, customer information, account information, and suspicious transaction records. Data preprocessing techniques, including cleaning, feature engineering, encoding, and class imbalance treatment, were applied to prepare the dataset for analysis. Three supervised machine learning algorithms which include Logistic Regression, Random Forest, and Gradient Boosting were developed and evaluated using performance metrics including ROC-AUC, Precision, Recall, F1-score, PR-AUC, and Matthews Correlation Coefficient. The findings showed that machine learning techniques can effectively improve fraud detection performance. Among the evaluated models, Random Forest achieved the best performance, with a ROC-AUC score of 0.9908, PR-AUC of 0.9305, recall of 95.6%, and F1-score of 0.6742, outperforming both the Logistic Regression and Gradient Boosting models. Feature importance analysis further revealed that customer age, account balance behaviour, and transaction patterns were among the strongest predictors of fraud. The study demonstrates that machine learning provides an effective and adaptive approach for strengthening fraud detection within microfinance institutions. The findings contribute practical evidence on the application of machine learning for fraud detection in Uganda’s microfinance sector and provide UGAFODE Microfinance with a potential framework for improving fraud risk management and transaction monitoring effectiveness.
Authentic leadership and employee resilience in health services
(Uganda Christian University, 2026-10-15) Angwech, Evalyne
The aim of the study was to investigate the relationship between authentic leadership and employee resilience in the healthcare sector. In Ugandan context, the study employed a cross-sectional survey design with a quantitative approach. Primary data was collected from a sample of 127 staff out of a population of 182 staff. The study employed both purposive and snowball sampling methods to select and reach respondents. A self-administered questionnaire with closed ended questions was used to collect data from respondents. The questionnaire passed both validity and reliability tests. Data was analyzed using SPSS software version 25. Descriptive and inferential statistics (correlations and regressions were performed. Correlation results revealed that; Self-Awareness, Internalized Processing, Relational Transparency, Balanced Processing significantly and positively associated with employee resilience. On the other hand, regression results revealed that; Self-Awareness, Internalized Processing, Relational Transparency and Balanced Processing are significant positive predictors of employee resilience, and they account for 37.1% (Adjusted R Square=0.371) variation in employee resilience. The study recommends that; Management of a health sector should design appropriate policies that can foster authentic leadership. These policies should cover areas such as internalized processing, relational transparency, balanced processing and self-awareness which can enhance employee to bounce back in case of any uncertainty. Management in health sector should also put emphasis on internalized processing, relational transparency, balanced processing and self-awareness as they are key predictor factors that can enhance employees continue with their operations in times of hardship and lastly, the ministry of health should support private hospitals to employee good leadership practices such as authentic leadership which in turn can enhance employee resilience. The study informs both policy and practical implication on authentic leadership and employee resilience during the covid-19 pandemics detailed in the subsequent chapters.
Interpretable spatial machine learning framework for evaluating malaria intervention effectiveness in Northern Uganda
(Uganda Chrisitian University, 2026-10-14) Sseguya, Henry
Malaria prevalence among children under five in Northern Uganda remained high in 2024-25 (37.4%), similar to 2018-19 (35.6%), despite a 13.2 percentage point rise in insecticide-treated net (ITN) coverage. This study examined why prevalence stagnated while coverage increased. It used Malaria Indicator Survey (MIS) data on 3,627 children in 110 clusters in Northern Uganda. The methods were wave-interaction logistic regression, Random Forest classification with SHAP value decomposition, and Local Indicators of Spatial Association (LISA). The association between ITN use and malaria reversed between waves (OR: 0.72 to 1.56; interaction OR = 2.38, p< 0.001). This reversal is better explained by confounding by indication, pyrethroid resistance and net degradation than by a real loss of protective effect. There was no significant wave interaction for indoor residual spraying (IRS) (OR = 1.03, p = 0.903). SHAP analysis identified a non-linear coverage threshold: the largest protective differences were associated with ITN coverage of 12% to 15%, below which population-level protection did not occur. Older children within the under-five group were the most affected, with the odds of malaria rising with each month of age in both waves. Prevalence differed sharply between sub-regions, from 14.9% in West Nile to 56.8% in Karamoja, and malaria was spatially clustered in both waves (Moran’s I = 0.28 and 0.31, p = 0.001). Hotspots persisted even though their ITN coverage was higher than in cold spots, showing that coverage alone has not reduced the burden where it is greatest. Successful malaria control in Northern Uganda requires a shift from aggregate coverage targets to spatially stratified, hotspot-based ITN distribution. PBO net scale-up should be accelerated in pyrethroid-resistant areas. IRS structural coverage should reach at least 85 percent to achieve community-level impact. Spatial surveillance should be built into national health information systems. These measures should be implemented as a package rather than as separate vertical programmes, because coverage gains are not enough when resistance and degradation reduce net efficacy. NMCP distribution should be coordinated using evidence from repeated MIS waves, so that programmatic decisions are based on sub-regional burden data.