Acute coronary syndrome risk prediction by ensemble‐MLPs

Abstract

Abstract Acute coronary syndrome (ACS) is a serious cardiovascular disease. The ACS risk prediction model is of great significance during the hospitalisation of ACS patients. However, traditional machine learning methods are not effective in predicting risk events in the ACS treatment process because of sample imbalance and noise. In this letter, the multilayer perceptron (MLP) and ensemble method are combined and then ensemble‐MLPs are proposed, which has made two innovations: 1) increase the diversity of the base MLP classifier at the data level, structure level, and parameter level and 2) improve the ensemble performance by proposing a new ensemble method using f1‐score weighted average. Experiments have shown that the proposed method outperforms conventional ensemble MLP method and other traditional machine learning methods on the task of predicting risk events in the ACS treatment process.