ABSTRACT
Objective:Rapid assessment of the outcome after percutaneous coronary intervention (PCI) in patients with acute coronary syndrome (ACS) is an important clinical issue. In this study, an electrocardiogram (ECG) analysis method based on dynamic learning was proposed.Methods:A total of 203 patients with ACS after successful PCI were enrolled for prospective analysis at the Emergency Department of Qilu Hospital of Shandong University from April 2019 to December 2020. All patients were divided into group without ≥70% postoperative stenosis ( n=72) and group with ≥ 70% postoperative stenosis ( n=131) according to the presence of 70% or more stenosis after PCI. The clinical data of ACS patients were collected and analyzed by χ2 test, t-test, or Mann-Whitney test. ECGs were recorded before and 2 h after PCI, and were dynamically analyzed to generate cardiodynamicsgram (CDG) using dynamic learning. In the group without ≥ 70% postoperative stenosis, the model and CDG index for evaluating myocardial ischemia were obtained by training support vector machine (SVM) using 10 times 10-fold cross-validation. Results:There was no significant difference in clinical data between the two groups. The prediction accuracy and sensitivity of the support vector machine model for myocardial ischemia in group without≥70% postoperative stenosis were 73.61%, and 84.72% respectively. CDG transformed from disorderly to regular after PCI, and CDG index decreased significantly ( P<0.001): 90.28% (65) patients in group without≥70% postoperative stenosis, and 79.39% (104) patients in group with≥70% postoperative stenosis had lower CDG indexes than before PCI. Conclusions:In this study, CDG obtained by dynamic learning can intuitively and effectively evaluate the changes of myocardial ischemia before and after PCI, which is helpful to assist clinicians to formulate the next treatment plan.