Machine learning model identified patients at high risk of ACS. (A) Variable importance analysis showing the variables from the most to the least important to influence risk of ACS. The risk of ACS was significantly higher in the high-risk group compared with that in the low-risk group in both training (B) and validation (C) cohorts. (D) Cumulative incidence of ACS is not significantly different in patients with and without ASXL1, DNMT3A, TET2, and JAK2 somatic mutations. CAD, coronary artery disease; CCF, congestive cardiac failure; MCV, mean corpuscular volume; IPSS-R, Revised International Prognostic Scoring System.