We aimed to evaluate the association between MASLD and major adverse cardiovascular events (MACE) in young Korean men and explore high-risk population using causal machine learning.

We conducted a nationwide cohort study using linked data from the Military Manpower Administration conscription examination, which includes nearly all Korean men, and the National Health Insurance claims database (2015–2024). MASLD was defined as hepatic steatosis index ≥ 36 with at least one cardiometabolic risk factor, including obesity (BMI ≥23 kg/m²), hypertension (blood pressure ≥ 130/85 mmHg or antihypertensives), diabetes (FPG ≥100 mg/dL or antidiabetics), and dyslipidemia (TG ≥150 mg/dL or lipid-lowering agent) The primary outcome was MACE, defined as myocardial infarction, stroke, heart failure, or cardiovascular death. After 1:1 propensity score matching, Cox PH models were estimated the association between MASLD and MACE. A Causal Survival Forest using 43 covariates was applied to estimate heterogeneous treatment effects. 198,184 MASLD and 1,245,126 non-MASLD male were included. Hazard ratio was 1.33 (95% CI, 1.29-1.41). The CSF estimated an average survival probability difference of -0.0030 (95% CI -0.0042 to -0.0017). Heterogeneous treatment effect was observed in covariates such as history of COPD. MASLD was associated with an increased risk of MACE in young Korean men. Causal machine learning identified clinically relevant high-risk subgroups, supporting targeted prevention strategies.