FACULTY OF COMPUTING AND SOFTWARE ENGINEERING SCHOOL OF POST-GRADUATE STUDIES MASTERS THESIS

Abstract

Hepatitis B is a major global public health concern. Hepatitis B is a viral infection that attacks the liver and can cause both acute and chronic disease. WHO estimates that 254 million people were living with chronic hepatitis B infection in 2022, with 1.2 million new infections and 1.1 million deaths, mostly from (primary liver cancer) each year. Ethiopia has a high incidence of viral hepatitis, as the study from 2022 states, with an anticipated 4.4 million cases of hepatitis B and 1.2 million cases of hepatitis C. The Federal Ministry of Health stated in the report 2020 Ethiopia lies in an epidemiologically defined region where the prevalence of Hepatitis B infection is classified as hyperendemic, with an estimated prevalence of 8–12%, while the prevalence of Hepatitis C infection is expected to be at least 2.5%. In Ethiopia, hepatitis is a serious health issue, with the Southern Omo Zone being among the worst afflicted. The most common causes of liver cirrhosis and liver cancer in the nation are hepatitis B and C. The purpose of this study is to use deep learning techniques to predict the spread of hepatitis B among pregnant women in the Southern Omo Zone Ethiopia. LSTM, GRU, and hybrid (ARIMA – LSTM) deep learning algorithms are used in the suggested solution. Model the integration of conventional time-series techniques and deep learning to forecast the spread of hepatitis B based on a range of demographic and socioeconomic variables. The study also involves collecting and analyzing data from Southern Omo zone Health office. The results of this study could help healthcare professionals in the Zone to identify whether the prevalence of hepatitis is high in the future, and to take proactive measures to prevent the spread of Hepatitis B. To choose a forecasting method, two deep learning algorithms and Hybrid model with traditional time-series methods are proposed to be explored and compared to find the best performance solution for Forecasting the spread of hepatitis B. The GRU model on rolling cross validation has the lowest error score and the highest accuracy in forecasting the number of incidence cases for HBV in both of HBV_Reactive and HBV_Reactive_Ratio. The researcher uses Python programming for data preparation, analysis, and model development, and E-draw max for drawing diagrams.