Machine Learning Identification of Patients with Non-ST Segment Elevation Acute Coronary Syndrome Using High-resolution Magnetocardiography

The aim of this study was to develop an automated assessment of non-elevated acute coronary syndromes (ACS). Major field-like, time-domain and image-like features that could be explained by electrophysiological mechanisms were extracted from the study population for modeling. A total of 723 cases of magnetocardiography data were selected for 50% cross-validation, and the clinical information of the patients in the validation set was evaluated by professional doctors. All patients underwent coronary angiography. The results show that the accuracy of MCG is 86.6%, the precision is 89.8%, the sensitivity is 81.9%, the specificity is 91.1%, the F1 score is 85.6%, the area under ROC curve is 0.9338, the sensitivity is 69.4%, the specificity is 73.2%. The results of this study indicate that it has important research and application value for ACS patients, especially those patients whose ECG does not indicate typical ST segment changes. The magnetocardiodiagnostic method proposed in this study provides a rapid and accurate diagnostic tool for clinicians and improves the interpretability of MCG data. In addition, some of the features revealed in this study may be related to the location of ischemia, providing an opportunity for further research.