IVARA,, DAVID FREDRICK (2026) A MACHINE LEARNING APPROACH TO CLINICAL DIAGNOSIS OF PEPTIC ULCER. Other thesis, Godfrey Okoye University, Enugu Nigeria.
|
Text
Ivara, David Fredrick.pdf Download (3MB) | Preview |
Abstract
The rate of Peptic ulcer disease (PUD) has been a common gastrointestinal ailment most times usually caused by H. pylori infection or regular usage of NSAIDs. Due to high costs and inaccessibility of existing diagnostic procedures, such as endoscopy and laboratory tests, PepticSense—an aided machine learning-based clinical decision support web application (CDSS) for preliminary detection of presence or absence of gastric or duodenal ulcers—is developed in this paper. The data was synthesized from 6,000 patients' records in Nigeria with 42 features representing demographic, social and lifestyle information, patient's health condition, medical history and lab results. In order to develop the solution in a structured way using the Agile Software Development method is discussed alongside the Object-Oriented Analysis and Design approach and Decision Tree, Support Vector Machine, Naive Bayes, and Random Forest classifiers, CRISP-DM process, data was cleaned, prepared, feature-engineered, encoded, selected, scaled, and split into train and test sets. Comparisons of Accuracy, Precision, Recall, F1 Score, and AUC-ROC have been evaluated based on their performance, models showed that the latter is the most accurate classifier. The classifier was hyperparameter tuned through GridSearchCV and incorporated into Django REST Framework application with Angular frontend, PostgreSQL database, JWT authentication and role-based access control system. After evaluation of its performance on 1,200 records from the same distribution, but not seen during development of the model, high metrics of accuracy, precision, recall, F1-score, and AUC-ROC and low misclassifications between similar ulcer types were achieved. Thus, PepticSense became an inexpensive solution that could be used in any healthcare environment.
| Item Type: | Thesis (Other) |
|---|---|
| Subjects: | Q Science > Q Science (General) |
| Divisions: | Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science |
| Depositing User: | Cynthia Ugwuoti |
| Date Deposited: | 22 Jul 2026 15:07 |
| Last Modified: | 22 Jul 2026 15:07 |
| URI: | http://eprints.gouni.edu.ng/id/eprint/6060 |
Actions (login required)
 |
View Item |
