ONBLR: An effective optimized ensemble ML approach for classifying liver cirrhosis disease

Abstract

Liver cirrhosis is a late stage of scarring (fibrosis) of the liver caused by many forms of liver diseases and conditions, such as hepatitis and chronic alcoholism. Early diagnosis and management can help slow the progression of the disease and improve quality of life. Developing more accurate and non-invasive methods for the early detection of liver cirrhosis to enhance timely intervention and improve patient outcomes. Effective prediction reduces overall cirrhosis patients' mortality due to various bleeding, making further study in this field a significant task. Therefore, we propose a novel ensemble machine learning based approach for classification. Label encoding and data normalization using the min-max approach are performed in preprocessing. Then, characteristics such as age, gender, Liver function tests (AST, ALT, alkaline phosphatase, bilirubin) and Medical history and comorbidities are retrieved using the ConvNeXt approach. Using the Improved Grasshopper optimization algorithm (IGOA), the essential features that increase the accuracy even more were selected. Finally, the optimized ensemble machine learning approach naïve Bayes and logistic regression (ONBLR) is employed to classify the liver cirrhosis disease. Harris Hawks optimization algorithm is employed to optimize the hyperparameters. The proposed approach is compared with the existing state-of-the-art machine learning approaches. While compared with existing approaches, the proposed model yields 99.18 % accuracy, 99.12 % of Sensitivity and 98.92 % of specificity which is greater among other approaches.