Deep generative models for physiological signals: A systematic literature review.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Physiological signals (ECG, EEG, PPG, EMG) are crucial for health monitoring.
• Deep generative models offer novel approaches for analyzing and synthesizing these signals.
• Existing reviews lack a comprehensive summary of state-of-the-art deep generative models in this domain.

Purpose of the Study:

• To provide a systematic literature review of deep generative models for physiological signals.
• To summarize the recent state-of-the-art deep generative models.
• To analyze applications, challenges, and evaluation methodologies.

Main Methods:

• Systematic literature search and review.
• Analysis of recent research on deep generative models for ECG, EEG, PPG, and EMG.
• Categorization of models based on architecture, applications, and datasets.

Main Results:

• Identification and summary of leading deep generative models for physiological signal analysis.
• Overview of key applications, including signal synthesis, anomaly detection, and data augmentation.
• Analysis of common challenges, such as data scarcity and model interpretability.

Conclusions:

• Deep generative models show significant promise for advancing physiological signal analysis.
• Standardized evaluation protocols and benchmark datasets are essential for future research.
• This review offers a foundational resource for researchers and practitioners in the field.