Prediction of Liver Disease Using Deep Learning and Artificial Neural Networks

Abstract

Abstract— Liver disease is a major cause of illness and death worldwide, and early diagnosis remains a challenge due to the limited availability of trained medical professionals for interpreting diagnostic tests. This paper presents a deep-learning-based approach for the prediction of liver disease using the Indian Liver Patient Dataset (ILPD), obtained from Kaggle and the UCI Machine Learning Repository, consisting of 583 patient records. Nine clinical attributes — age, sex, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, albumin, total protein, albumin/globulin ratio, and alkaline phosphatase — were used to train and evaluate several classification models, namely Logistic Regression, Random Forest, and an Artificial Neural Network (ANN). The dataset was preprocessed to remove missing and duplicate values, encode categorical attributes, and normalize numerical features before being split into training (80%) and testing (20%) subsets. Feature-importance analysis identified ALT as the most influential predictor of liver disease and alkaline phosphatase as the least significant. The trained model was deployed through a Flask-based web application that allows users to input clinical parameters and receive an instant prediction. Experimental results demonstrate that the proposed system can effectively assist in the early screening of liver disease, offering a low-cost, accessible, and user-friendly diagnostic aid that bridges the gap between patients and healthcare providers.