Predictive Modeling for Mortality Risk in Nonalcoholic Fatty Liver Disease Patients: A Machine Learning Approach

This study explored the use of machine learning models to predict mortality risk among patients diagnosed with nonalcoholic fatty liver disease (NAFLD) using retrospective data spanning from 1997 to 2014. The dataset included data from 17,549 adult participants diagnosed with NAFLD and included various demographic, anthropometric, and biochemical parameters for analysis. Six machine learning algorithms—Decision Tree, Random Forest, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Logistic Regression, and AdaBoost Classifier—are evaluated for mortality risk prediction. AdaBoost has emerged as the most promising model, achieving near-perfect accuracy (99%) and high precision, recall, and F1-scores. Notably, the robust performance of AdaBoost suggests its potential for accurately predicting mortality risk among NAFLD patients. However, weaker performance is observed in models such as KNN, support vector classifier, and logistic regression. This study underscores the significance of machine learning in enhancing risk prediction and personalized management approaches for NAFLD patients.