International Research Journal of Education and Technology

Hepatitis is a liver illness that can be life- threatening if not diagnosed and treated in a timely manner. The severity of the disease must be graded with precision so as to inform the right treatment for better patient outcomes. This study examines the use of machine learning agents- K-Nearest Neighbors (KNN), Convolutional Neural Networks (CNN), and Logistic Regression (LR) – to classify patients depending on the severity of hepatitis. According to the dataset that included the clinical diagnosis, blood tests, and patient characteristics, the models acquired were used to separate the ascribed cases of hepatitis into; mild, moderate, and severe ones. K-Nearest Neighbors was used due to its ease in pattern searching on the feature space through proximity, CNN on the other hand due to its capability of finding important features in complex data without supervision, that conduct enhancement to LR which is mainly practiced for two categories. The models were assessed on the parameters of accuracy, precision, recall, and F1-score. It was found in the initial research that CNN outperformed KNN and LR in the classification of the severity of hepatitis infection with the effective use of complex information about the disease, which is a rare ability. KNN and LR also appeared to be competitive immediately after diagnosing even early lesions. This study shows the power of machine learning, and more specifically CNN, increasing the accuracy of grading hepatitis severity, providing an opportunity for physicians to enhance their diagnostic accuracy and the management of their patients.

Keywords— Hepatitis, Severity Grading, Machine Learning, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Logistic Regression (LR), Medical Diagnostics, Disease Prediction, Liver Disease Classification, Clinical Features, etc.