This study aims to use machine learning technology to construct an efficient fatty liver prediction model to help medical personnel accurately identify and classify high-risk populations for fatty liver, thereby providing support for early intervention and personalized treatment.

The research data was sourced from the National Health and Nutrition Examination Survey (NHANES) database in the United States from 2017 to 2020. Finally, 11 variables were included for analysis. Using 6 machine learning algorithms, Decision Tree (DT) Extreme Gradient Boosting (XGBoost), Adaptive Boosting (Adaboost), K-Nearest Neighbor (KNN), Logistic Regression, and Artificial Neural Network (ANN) models are used to construct a fatty liver prediction model. The randomly generated 70% is used as the training set, and the remaining 30% is used as the testing set. The model evaluation metrics include accuracy, precision, recall, F1 score, Receiver Operating Characteristic curve (ROC), and Area Under Curve (AUC).

A total of 3383 participants were included, including 1506 patients with fatty liver. The decision tree The AUC of XGBoost, AdaBoost, KNN, logistic regression, and ANN models are 0.98838, 0.94624, 0.88935, 0.99110, 0.90047, and 0.98990, respectively. Except for the KNN model, the calibration curves of all other models performed well, with the AdaBoost model showing the best predictive performance among the six models.

The AdaBoost model is considered the optimal predictive model for predicting fatty liver and can explain the interaction between feature variables. Body mass index, triglycerides, and low-density lipoprotein indicators can effectively identify high-risk populations for fatty liver.