School of Computing and Informatics Technology (CIT) Collection
Permanent URI for this collection
Browse
Recent Submissions
1 - 5 of 565
-
ItemModel for predicting the level of productivity of Cowpea (Vigna Uguiculata) sprout germination using computer vision(Makerere University, 2025)This dissertation explores the impact of estimating the level of productivity of cowpea (vigna unguiculata) germination using computer vision. The primary objective was to develop a model for monitoring and estimating the level of germination in vigna crops using computer vision in Unyama Municipality in Gulu Uganda. Using experimental, machine learning methods of Convolutional Neural Networks (CNN), and ResNet50 FPN, the study investigated germination. The findings indicate that the development, training, and testing of a deep-learning model for estimating levels of vigna unguiculata (cowpeas). The primary dataset for model training and validation was obtained from a controlled experimental greenhouse environment. An open-source dataset from Kaggle was used to evaluate the accuracy of the object detection model over three epochs. Intersection over union (IoU) match of 100% was reported, indicating a perfect match. A sprout success rate 64% based on leaf count was reported. The study contributes to artificial intelligence in the field of computer vision by Contributed to the field of knowledge, creation of a vigna dataset, development of a vigna detection model, detect and count the number of vigna, identification and classification of objects within the dataset. Keywords: ResNet50 FPN, Convolutional Neural Networks (CNN), Germination.
-
ItemA predictive model for estimating students’ final cumulative GPA at graduation a case study of Makerere University(Makerere University, 2025)Accurate prediction of students’ final Cumulative Grade Point Average (CGPA) at graduation is crucial for identifying at-risk students and improving academic outcomes. This study focuses on developing a predictive model for estimating CGPA/Degree Class using machine learning techniques, leveraging a dataset of over 2500 records collected from Makerere University graduates. The dataset encompasses academic performance, demographic details, and socioeconomic factors, mapped to official university records to ensure accuracy and credibility. Multiple machine learning models, including Logistic Regression, Gradient Boosting Classifier, and Random Forest Classifier, Dummy Classifier, Extra Trees Classifier, Neural Network MLPClassifier and Decision Tree Classifier were evaluated across three training-test splits (80/20, 75/25, and 70/30). Gradient Boosting Classifier emerged as the best-performing model, achieving consistent accuracy of approximately 84.7% and demonstrating robustness and interpretability. Feature importance analysis highlighted prior academic performance, students behaviour, and Alevel Year as the most significant predictors of CGPA, aligning with global findings and offering localized insights specific to Makerere University. Despite its contributions, the study faced challenges such as dataset size constraints and limited exploration of interdisciplinary variations. Nonetheless, the results underscore the potential of machine learning in academic performance prediction, providing actionable insights for educational institutions. The findings emphasize the importance of early intervention, targeted support, and data-driven decision-making in improving graduation rates. This research advances predictive analytics in education, offering a scalable framework for institutions. Future work should expand dataset size and include discipline-specific predictors for deeper analysis.
-
ItemResponsible disease modeling and prediction of Cardiovascular diseases(Makerere University, 2025)The increased number of deaths of cardiovascular diseases among people in both Low- and Middle-Income countries and in developed countries is alarming. There are Machine Learning (ML) models that have been developed for early diagnosis of cardiovascular diseases, however, their success is low due to the black box nature of the models and the trust among the cardiovascular diseases’ health experts is low thus hindering the models’ acceptance. This research study focused on developing explainable AI models to predict the likelihood of acquiring CVDs over a period of ten years from which the best performing one will be chosen. The open cardiovascular disease study dataset was used to develop multiple Machine learning models i.e K-Nearest Neighbors (KNN), Logistic Regression, XGBoost, Catboost, Random Forest, Naive Bayes, Ada Boost, Support Vector Machine, Gradient Boosting Machine, Long Short-Term Memory and Decision tree. The models’ performance was assessed using F1-score, Accuracy, Area Under the Curve, Precision, Recall, sensitivity, specificity and the confusion matrix metrics. From this study, it is observed that the Random Forest model performs better than the other models with an accuracy of 98% followed by XGBoost with 89% and KNN with 88%. Explainable AI techniques (XAI) like SHapley Additive exPlanations (SHAP) explainable technique, Partial Dependence Plots (PDP), Individual Conditional Expectations (ICE) and Local Interpretable Model-agnostic Explanations (LIME), were later applied to all the models to understand how they came to their prediction thus breaking the black box nature of Machine learning models. This research contributes to the identification of cardiovascular diseases risk factors with the use of feature learning and XAI for the early diagnosis of cardiovascular diseases thus aiding in early intervention. The leading risk factors that were established as per the models’ predictions are Age,sex, systolic Blood pressure(SysBP) and Cigarettes per day whereas diabetes, total cholesterol, Blood Pressure Medication (BPMeds) and prevalent Stroke are the least contributing risk factors implying they are not as important in acquiring CVDs.
-
ItemExplainable machine learning for viral rebound prediction in patients on antiretroviral therapy(Makerere University, 2025)HIV continues to be a global public health concern, as shown by the approximately 2.1 million HIV-positive patients who were receiving ART without experiencing viral suppression in 2022, according to UNAIDS. Although ART has significantly reduced HIV-related morbidity and mortality, viral rebound (VR) continues to threaten treatment success. In low-resource settings like Uganda, clinical systems have limited data to identify high-risk patients before VR occurs, which increases the likelihood of HIV transmission, treatment failure, drug resistance, and other challenges. Examining VR feature interaction patterns by healthcare workers is challenging due to complex interplay of factors associated with VR. Previous works on VR contributing factors have used less interpretable methods that cannot be trusted and adopted by healthcare workers. Demographic, clinical and laboratory data was preprocessed and divided into chunks of 70%, 15%, and 15% in order to train, validate, and test the six ML models (Hard Voting, Gradient Boosting, LightGBM, Random Forest, XGBoost, and Stacking), respectively. To address class imbalance, SMOTE was applied to the training data, before ML models were trained. Although Hard Voting achieved the highest F1-score (51.56), LightGBM (F1-score of 50.83) was selected as the best-performing ML model due to its superior interpretability when paired with XAI techniques (SHAP, LIME, ELI5, and PDP). Key predictors included past two or more consecutive suppressions, regimen history, adherence patterns, and ART duration. This study demonstrates that ML can provide accurate predictions of VR in ART patients, that can aid in the efficient allocation of healthcare resources by identifying which patients are most at-risk. VR feature interaction patterns will aid in creation of a valuable clinical decision support tool that will notify healthcare workers of patients that need tailored therapies and devise measures to promote viral suppression. XAI techniques integrated ensure that medical professionals understand prediction results and foster proper trust, ultimately strengthening HIV care.
-
ItemExplainable machine learning for viral rebound prediction in patients on antiretroviral therapy(Makerere University, 2025)HIV continues to be a global public health concern, as shown by the approximately 2.1 million HIV-positive patients who were receiving ART without experiencing viral suppression in 2022, according to UNAIDS. Although ART has significantly reduced HIV-related morbidity and mortality, viral rebound (VR) continues to threaten treatment success. In low-resource settings like Uganda, clinical systems have limited data to identify high-risk patients before VR occurs, which increases the likelihood of HIV transmission, treatment failure, drug resistance, and other challenges. Examining VR feature interaction patterns by healthcare workers is challenging due to complex interplay of factors associated with VR. Previous works on VR contributing factors have used less interpretable methods that cannot be trusted and adopted by healthcare workers. Demographic, clinical and laboratory data was preprocessed and divided into chunks of 70%, 15%, and 15% in order to train, validate, and test the six ML models (Hard Voting, Gradient Boosting, LightGBM, Random Forest, XGBoost, and Stacking), respectively. To address class imbalance, SMOTE was applied to the training data, before ML models were trained. Although Hard Voting achieved the highest F1-score (51.56), LightGBM (F1-score of 50.83) was selected as the best-performing ML model due to its superior interpretability when paired with XAI techniques (SHAP, LIME, ELI5, and PDP). Key predictors included past two or more consecutive suppressions, regimen history, adherence patterns, and ART duration. This study demonstrates that ML can provide accurate predictions of VR in ART patients, that can aid in the efficient allocation of healthcare resources by identifying which patients are most at-risk. VR feature interaction patterns will aid in creation of a valuable clinical decision support tool that will notify healthcare workers of patients that need tailored therapies and devise measures to promote viral suppression. XAI techniques integrated ensure that medical professionals understand prediction results and foster proper trust, ultimately strengthening HIV care.