A Systematic Survey of Data Augmentation of ECG Signals for AI Applications

Area of Science:

• Cardiology
• Artificial Intelligence
• Signal Processing

Background:

• Artificial intelligence (AI) models for electrocardiogram (ECG) analysis require large labeled datasets, which are difficult to obtain.
• Data augmentation (DA) strategies offer a solution to improve AI model performance by artificially expanding datasets.
• The effectiveness and application of DA in ECG analysis remain areas needing comprehensive understanding.

Approach:

• A systematic literature review following PRISMA guidelines was conducted.
• Publications from 2013-2023 were searched across major databases (IEEE Explore, PubMed, Web of Science).
• 119 relevant papers were analyzed and categorized by AI application, leads, DA method, classifier, performance gains, and datasets.

Key Points:

• DA methods show potential in improving AI-based ECG analysis performance.
• Categorization reveals trends in DA techniques and their impact across various ECG applications.
• The review synthesizes current knowledge on DA for ECG signals, identifying research gaps and future directions.

Conclusions:

• Data augmentation is a promising technique for advancing AI in ECG diagnosis and monitoring.
• Further research into optimized DA strategies can unlock greater potential for AI in cardiology.
• This review provides a foundational understanding for researchers and developers in the field of AI-powered ECG analysis.