Prediction of longitudinal clinical outcomes after acute myocardial infarction using a dynamic machine learning algorithm

• DOI: 10.3389/fcvm.2024.1340022
• Publication Date: 2024-04-05T14:18:12Z
• Authors: Joo Hee Jeong; Kwang-Sig Lee; Seong-Mi Park; So Ree Kim; Mi-Na Kim; Shung Chull Chae; Seung-Ho Hur; In Whan Seong; Seok Kyu Oh; Tae Hoon Ahn; Myung Ho Jeong
• Keywords: Mace

Several regression-based models for predicting outcomes after acute myocardial infarction (AMI) have been developed. However, prediction that encompass diverse patient-related factors over time are limited. This study aimed to develop a machine learning-based model predict longitudinal AMI. was based on nationwide prospective registry of AMI in Korea ( n = 13,104). Seventy-seven predictor candidates from prehospitalization 1 year follow-up were included, and six learning approaches analyzed. Primary outcome defined as 1-year all-cause death. Secondary included deaths, cardiovascular major adverse event (MACE) at the 3-year follow-ups. Random forest resulted best performance primary outcome, exhibiting 99.6% accuracy along with an area under receiver-operating characteristic curve 0.874. Top 10 predictors peak troponin-I (variable importance value 0.048), in-hospital duration (0.047), total cholesterol maintenance antiplatelet (0.045), coronary lesion classification (0.043), N-terminal pro-brain natriuretic peptide levels (0.039), body mass index (BMI) (0.037), door-to-balloon (0.035), vascular approach (0.033), use glycoprotein IIb/IIIa inhibitor (0.032). Notably, BMI identified one most important revealed distinct effects each highlighting U-shaped influence MACE Diverse time-dependent variables postdischarge period influenced Understanding complexity dynamic associations risk may facilitate clinical interventions patients