A Review on Diagnosing Liver Diseases Based on Machine and Deep Learning Algorithms

Abstract

This study aims to provide an overview of the studies on supported machine learning and deep learning methods for predicting liver disease. Because of the liver's crucial role in filtering blood and conducting essential activities, liver disease is a severe medical illness that can be fatal. By examining historical datasets, machines and deep learning algorithms provide a means of making early predictions of liver disease. To better understand dataset features, this work aims to investigate the usage of supervised machines and deep learning algorithms for liver disease prediction the research focuses on leveraging historical data sets, such as the Indian Liver Patient Dataset (ILPD), BUPA Dataset and Youden’s Index were used for the investigations. Various methods such as Logistic Regression, Sequential Minimal Optimization model, Random Forest algorithm, K-Nearest Neighbors, Gaussian (SVM) model, Support Vector Machines (SVM), RNN, Bidirectional Long Short-Term Memory (BiLSTM), GBM model, Artificial Neural Networks (ANN) and others are used. The accuracy of the findings is greater than 70% for diagnosing liver disease. Pre-processing methods such as univariate and bivariate analysis are also employed that have accuracy that is less than from the other machine learning methods. The study assesses the efficacy of machine and deep learning algorithms in correctly predicting liver disease by building a multi-class classification model and utilizing metrics. The suggestions obtained from the results demonstrate the promise of machine and deep learning in healthcare and emphasize the significance of ongoing research in constructing sophisticated and accurate prediction models for diagnosing liver disease.