AUTHOR=Li Zhengliang , Chen Xiaokai , Ren Linlin , Wang Banghui , Ruan Shimiao , Wang Juan , Zhang Wenzhong 

TITLE=Explainable machine learning model for classifying atherosclerotic cardiovascular disease in patients with metabolic dysfunction-associated steatotic liver disease

JOURNAL=Frontiers in Endocrinology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2025.1684558

DOI=10.3389/fendo.2025.1684558

ISSN=1664-2392

ABSTRACT=BackgroundCardiovascular disease (CVD) is the leading cause of mortality in patients with metabolic dysfunction-associated steatotic liver disease (MASLD), yet traditional risk predictors remain limited in clinical practice.ObjectiveTo develop machine learning (ML) models for classifying prevalent atherosclerotic cardiovascular disease (ASCVD) risk in MASLD patients, and to enhance model interpretability using SHapley Additive exPlanations (SHAP). Methods: This retrospective study included 590 MASLD patients diagnosed at the Affiliated Hospital of Qingdao University between December 2019 and December 2024. Patients were randomly divided into a training set (n=413) and a validation set (n=177), and further stratified based on ASCVD status. Least absolute shrinkage and selection operator (LASSO) regression was used for feature selection. Six ML models were developed and evaluated using sensitivity, specificity, accuracy, area under the receiver operating characteristic curve (AUC), and F1 score. SHAP analysis was performed to interpret feature contributions.ResultsASCVD was present in 434 of 590 patients (73.6%). The Gradient Boosting (GB) model achieved the best performance, with AUCs of 0.918 (95% CI: 0.890–0.944) in the training set and 0.817 (95% CI: 0.739–0.883) in the validation set. SHAP analysis identified the top predictors as the Cholesterol–HDL–Glucose (CHG) index, Castelli Risk Index II (CRI-II), lipoprotein(a) [Lp(a)], serum creatinine (Scr), and uric acid (UA).ConclusionThe GB model demonstrated strong high accuracy in identifying existing ASCVD in MASLD patients and may serve as a useful tool for early risk stratification in clinical settings.