AI and Machine Learning Applications in Hepatitis B Diagnosis: An Exploration of Techniques and Outcomes

Hepatitis B is one of the major global health challenges, affecting millions worldwide and frequently leading to serious liver conditions like hepatocellular carcinoma and cirrhosis. Early and accurate diagnosis is very important for effective management and treatment; however, most of the traditional diagnostic approaches, including serological tests, liver function tests, liver biopsy, etc., are very often affected by limitations in resource availability and diagnostic accuracy. This review paper highlights the current development in various research works on the application of artificial intelligence (AI) and machine learning (ML) in the diagnosis of Hepatitis B. This paper reviews all the major ML computational techniques used in Hepatitis B diagnosis, underlining the use of various computational techniques: Generalized regression neural networks (GRNN), multilayer Mamdani fuzzy inference system (ML-MFIS), hybrid models combining clustering with ensemble learning, ANFIS, deep neural network (DNN), and support vector machine (SVM). These methodologies have been reported to show substantial improvement in the performance of diagnosis, where the maximum accuracy rate is up to 99.35%. This paper has also listed the datasets used for the studies and compared the performance metrics of these studies in terms of precision, accuracy, and F-score. Despite the progress, several issues persist, such as small dataset sizes, data imbalance, and the need for thorough validation across diverse clinical settings. Several research questions are also outlined to showcase the current ML diagnosis methods and their performances along with their limitations to provide the future research direction. It also covers the adoption of explainable AI (XAI) as of one the potential future aspect.