A Hybrid Multilayer Stack Ensemble Model for Early Prediction of Liver Disease

Abstract

One of the leading causes of death worldwide is liver disease. The number of people experiencing suffering is increasing consistently. Maintaining a healthy liver is crucial for supporting essential processes, including digestion and detoxification. Some of the most common liver issues that need medical treatment include fatty liver, cirrhosis, and hepatitis. Due to the mild symptoms, it is challenging to anticipate in the early phases. To tackle the problem, various heterogeneous data mining algorithms are used to analyse performance and identify the most appropriate model for diagnosing liver disease. In our research, we utilised a multi-layered stacked ensemble learning model to enhance the precision of liver disease prediction. Five different classification models, including Support Vector Machine, Decision Tree, XG Boost, Cat Boost, and Logistic Regression models, were employed in the base layer of our model. The meta-layer of the model consists of K-Nearest Neighbour, Logistic Regression, Support Vector Machine classification models, and we classified the observations based on the Voting Classifier that we deploy on the meta-layer model. Consequently, we discovered that this suggested framework achieves a 92.35% accuracy rate along with improved F1 Score, recall, and precision.