Artificial Intelligence for Myocardial Infarction Detection via Electrocardiogram: A Scoping Review

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Diagnostics

Background:

• Acute myocardial infarction (MI) remains a leading cause of mortality globally, posing significant healthcare challenges.
• Early and accurate MI detection is crucial but often difficult despite diagnostic advancements.
• Artificial intelligence (AI) is emerging as a powerful tool to enhance electrocardiogram (ECG)-based MI detection.

Purpose of the Study:

• To systematically map and evaluate the applications of AI in detecting MI using ECG data.
• To provide a comprehensive overview of AI-driven MI detection methods.
• To identify trends and challenges in AI for ECG-based MI diagnosis.

Main Methods:

• A systematic scoping review of AI applications for MI detection via ECG.
• Searches conducted in major databases (MEDLINE, Embase, Web of Science, Cochrane) from 2015 to October 2024.
• Inclusion of 220 studies following PRISMA-ScR guidelines, extracting data on AI models, algorithms, ECG types, and performance metrics.

Main Results:

• AI applications for MI detection have rapidly increased since 2015, peaking in 2022.
• Convolutional neural networks and support vector machines were predominant AI models, often using 12-lead ECGs.
• High reported performance metrics were often based on small, single-source datasets with optimistic validation, showing limited generalizability and potential biases.

Conclusions:

• AI-based MI detection using ECGs is a rapidly growing field.
• Current diagnostic performance is significantly constrained by dataset limitations and validation methodologies.
• Standardization in reporting, datasets, and validation is essential for clinical integration, explainability, and equitable deployment of AI in MI detection.