Comparative Analysis of Liver Disease Prediction Using Machine Learning Models

Abstract

Liver disorders represent a major global health challenge and are responsible for a significant number of hospitalizations and long-term complications. Early identification of liver abnormalities is essential for improving treatment outcomes and reducing mortality. However, conventional diagnosis methods rely heavily on laboratory tests and expert interpretation, which are time-consuming and costly.

This paper presents a comparative study of machine learning algorithms for predicting liver disease using routinely collected clinical and biochemical parameters. Several supervised learning models, including Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbour, Gradient Boosting and Neural Network based models, are implemented and evaluated. The Indian Liver Patient Dataset is used as the benchmark dataset.

A complete data processing pipeline is developed, consisting of data cleaning, feature transformation, attribute ranking and model evaluation. The performance of the models is analysed using accuracy, precision, recall and F1-score. Experimental results demonstrate that ensemble and neural-based models provide superior predictive performance compared to traditional classifiers. The study confirms that machine learning can serve as a reliable decision-support tool for early liver disease screening.