Background: Myocardial infarction (MI) diagnosis relies on established clinical criteria, primarily involving the identification of electrocardiographic (ECG) abnormalities, such as ST-segment elevation in anatomically contiguous leads. Although rule-based algorithms grounded in these guidelines remain prevalent in clinical practice, recent advances in deep learning (DL) have demonstrated promising performance due to their capability to extract subtle patterns from raw ECG signals. However, there is a lack of studies comparing the performance of rule-based software with that of DL models.
Objectives: The primary objectives of this study were to develop a DL model and to evaluate its performance in comparison with a rule-based algorithm for MI detection, anatomical localization (anterior, lateral, inferior, and septal), and stadium classification (acute, chronic, and normal).
Methods: In this study, we developed a multitask-based DL model for three distinct tasks: MI detection, localization, and stage classification. By leveraging shared representations across tasks, the multitask learning framework enabled the model to capture common features and inter-task dependencies, thereby improving generalization and reducing computational costs. The model was trained using 12-lead median beats from the PTB-XL+ dataset, with 7855 patients included in the training set and 759 patients in the test set.
Results: The DL model achieved a sensitivity (Se) of 0.89 and a specificity (Sp) of 0.96 for MI detection, outperforming the rule-based algorithm, which achieved a sensitivity of 0.69 and a specificity of 0.94. For MI localization, the DL model achieved an average F1 score of 0.72, compared to 0.55 for the rule-based algorithm. In MI stadium classification, the DL model attained an average F1 score of 0.68, compared to 0.58 for the rule-based method. Overall, the DL model consistently outperformed the rule-based algorithm across all tasks, highlighting its potential as a more effective and flexible alternative for automated MI assessment.