As a clinically acute disease, AMI is often associated with arrhythmia, which complicates the patient's condition and increases the Incidence of adverse events (including stroke [27], higher use of pacemakers [2], re-infarction, cardiogenic shock, heart failure, asystole [5], and sudden cardiac death [28])At the same time, patients with arrhythmia have a significantly higher in-hospital mortality [2, 4, 27, 29], 30-day mortality [30, 31] and 1-year mortality[5] than those without arrhythmia. Therefore, it is essential to predict the occurrence of arrhythmia after AMI as early as possible. To this end, a large number of studies have analyzed the risk factors for arrhythmia after AMI [5, 7–9, 30, 32–37], but there is no systematic risk model. Currently, AMI's clinical risk model is mainly the GRACE risk score recommended by the ACC/AHA guidelines [38]. Still, it is mainly used to assess patients' mortality and may not accurately predict the occurrence of arrhythmia. Besides, the model is constructed using traditional statistical methods and only linearly analyzes the relationship between a few factors,do not address the potential prognostic value of interactions between several unexpected weaker risk factors and the primary outcome. For complex diseases, multi-factor and multi-level interactions need to be analyzed. In this case, ML can provide a useful alternative when encountering a large number of potentially relevant variables when building a predictive model. In the cardiovascular field, ML has been used in medical image analysis [39–42]༌disease classification and diagnosis [14, 17, 43, 44], and predictive model construction [19, 23, 26, 45, 46]. At present, researches related to ML and AMI were mainly devoted to the prediction of patient mortality [23, 47]༌and the ML model of arrhythmia after AMI has not been explored. In this study, we sought to harness the power of big data analytics and ML to develop an ML-based prediction model for the occurrence of tachyarrhythmia after AMI.
We applied 3 ML techniques (decision tree, random forest, ANN) to evaluate the risk of tachyarrhythmia after AMI. We found that the ANN algorithm's prediction ability in both the full variable model and after feature selection compared with other machine classifiers. After feature selection, the ANN model obtained the best prediction performance (accuracy of 0.654, AUC of 0.597). To evaluate our ML results clinically, we referred to the GRACE variables utilized in the models implemented in current AHA/ACC Guidelines, which are widely used for risk assessment in patients with AMI.
Our results show that the overall performance of ML was moderate, and therefore, it probably cannot yet replace diagnostic or risk estimations that further workup can provide. Nevertheless, when results were compared to those of utilizing the sets of variables considered in the Grace models, ML exhibited a higher performance for predicting the occurrence of tachyarrhythmia after AMI. Therefore, the ML-based prediction model can as a supplement to the current risk score.
Before ML, we included 45 variables based on the current AMI risk score [1,31,48−52] and the risk factors for tachyarrhythmia after AMI identified in previous studies [5–7,9,31–34,53−55]. We used the information gain method to select the top 15 highly predictive variables for the ML model construction to reduce the data dimension. Furthermore, Using ML feature-selection ranking, we found that certain risk factors that were not included in previous risk scores were significant predictors, such as angiographically determined lesion location, ultrasonic ventricular wall motion parameters abnormalities, right atrium diameter, and bundle-branch-block. These variables can be used as supplements to the current AMI risk score to assist clinicians in disease assessment.
Limitation:
The present study naturally carries the limitations of any observational study. However, this kind of largescale retrospective analysis is the main target of the data-driven approaches of ML. Second, this ML approach still needs further model training, validation, and optimization before clinical application. Patients in this study were enrolled from a single center that included only Chinese patients. Nevertheless, we compared the performance of advanced ML algorithms with the GRACE variable set model. The main finding of the current analysis was that ANN exhibited the highest prediction performance. ML-based prediction model could represent a great supplement in optimizing risk assessment and even clinical alerts of patients after AMI.