HepatoAI: A Machine Learning-Based Web Application for Multi-Stage Liver Cirrhosis Prediction

Abstract

Liver cirrhosis is a chronic, progressive liver disease that causes irreversible scarring of liver tissue and significantly impairs hepatic function. Early and accurate identification of the disease stage is essential for effective clinical intervention and improved patient outcomes. This paper presents HepatoAI, a machine learning-based web application designed to predict multi-stage liver cirrhosis from structured patient clinical data. Five classification algorithms were systematically evaluated: Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, and Gradient Boosting. Gradient Boosting achieved the highest classification accuracy of 53.01%, followed by Decision Tree (50.60%) and Random Forest (49.40%). The Random Forest model was selected for deployment owing to its superior feature-importance transparency and clinical interpretability. The application was built using Flask and is publicly accessible at https://vhshreya17-hepatoai.onrender.com. The system classifies patients into four clinically meaningful stages — Early Fibrosis, Moderate Fibrosis, Advanced Fibrosis, and Severe Cirrhosis — and provides confidence scores, risk assessments, and feature-importance visualizations. The proposed system demonstrates the practical value of machine learning in healthcare decision-support applications and establishes a foundation for further research in intelligent hepatic disease prediction.

Keywords