GIÁ TRỊ TIÊN LƯỢNG TỬ VONG CỦA MÔ HÌNH MÁY HỌC RỪNG NGẪU NHIÊN Ở BỆNH NHÂN NHỒI MÁU CƠ TIM CẤP

Abstract

Objectives: To evaluate the prognostic value of a random forest machine learning model for predicting mortality in patients with acute myocardial infarction (AMI). Subjects and methods: This retrospective cohort study included 540 patients with AMI treated at Cho Ray Hospital from January 2020 to September 2021. Clinical, laboratory, coronary angiographic, and treatment data were used to train a random forest model using Python 3.12. Data were standardized, class imbalance was addressed with the SMOTE method, and principal component analysis was applied to identify key predictors. Model performance was assessed using the area under the receiver operating characteristic curve (AUC) and accuracy. Results: The mean age was 64.0 ± 11.6 years, and 71.3% were male. The 12-month mortality rate was 10.7%. The most important predictors of mortality included Killip class, clinical type of AMI, lesions of the right coronary artery and the left anterior descending artery, smoking, use of angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers, anemia, admission troponin I level, age, and Gensini score. In the original dataset, the model achieved AUC values ranging from 0.653 to 0.730, with accuracy between 0.889 and 0.907. After data standardization and class balancing, AUC improved to 0.709–0.730. Conclusion: The random forest machine learning model demonstrated fair prognostic performance for mortality in patients with AMI.