Analysis code for: Machine Learning Prediction of Significant Liver Fibrosis in MASLD Using Routine Clinical Indicators (NHANES 2017–2020)
Description
Analysis code accompanying the manuscript "Machine Learning Prediction of Significant Liver Fibrosis in Metabolic Dysfunction-Associated Steatotic Liver Disease Using Routine Clinical Indicators: A Cross-Sectional Study Based on NHANES 2017–2020". The archive contains 14 scripts (12 Python, 2 R) covering data download, cohort construction, model development (logistic regression, random forest, XGBoost, LightGBM, SVM, MLP), hyperparameter tuning, probability calibration, decision curve analysis, SHAP interpretability, internal validation, subgroup fairness analysis, and figure/table generation. Source data are publicly available from NHANES ( https://www.cdc.gov/nchs/nhanes/) and are not redistributed; script 01_download.py retrieves them automatically. Environment: Python 3.12, R 4.6.1. Run scripts in numerical order. See README.md in the archive for details.
Files
Institutions
- First Affiliated Hospital of Xi'an Jiaotong UniversityShaanxi, Xi'an