Research on Electrocardiogram Abnormality Detection Method Based on Integrated Learning Models

For the early detection and prompt treatment of arrhythmias and other related problems, ECG anomaly detection is essential to the diagnostic process of cardiovascular illnesses. But for ECG anomaly detection applications, conventional single models frequently face drawbacks such overfitting and low robustness to abnormalities. To enhance the accuracy and robustness of ECG anomaly detection, this research emphasizes the application of ensemble learning techniques, particularly by constructing an ensemble model that employs the AdaBoost algorithm in conjunction with Random Forest as the base classifier. The experimental outcomes demonstrate that the ensemble model achieves outstanding performance in various assessment metrics, including precision, sensitivity, and F1 score. We validated our findings using the reputable MIT-BIH Arrhythmia Database. In particular, the model's overall accuracy was 99.53%, its recall rate for ventricular ectopic beats (V) was 99.57%, and its overall F1 score was 97.7%. The ensemble model suggested in this study exhibits notable benefits over conventional machine learning algorithms and deep methods of learning, offering strong technical support and a reference for clinical ECG diagnosis decision-making.