Exploring the Power of PPG Matrix for Atrial Fibrillation Detection with Integrated Explainability
Published in Engineering Applications of Artificial Intelligence, 2023
Recommended citation: Fuster-Barceló, C., Guerrero-López, A., Camara, C., & Peris-Lopez, P. (2024). Exploring the power of photoplethysmogram matrix for atrial fibrillation detection with integrated explainability. *Engineering Applications of Artificial Intelligence*, 133, 108325. https://doi.org/10.1016/j.engappai.2024.108325
Exploring the Power of PPG Matrix for Atrial Fibrillation Detection with Integrated Explainability
Published in: Engineering Applications of Artificial Intelligence
🧠 Summary
This study introduces a novel explainable framework for atrial fibrillation (AF) detection using Photoplethysmogram (PPG) signals. The key innovation lies in converting PPG data into PhotoplethysmoMatrices (PPMs)—heatmap-like visualizations of aligned cardiac peaks—and applying a CNN with attention modules for classification.
🧪 Results
- Achieved 100% classification accuracy across multiple experiments.
- Incorporated attention mechanisms for enhanced model interpretability.
- Showed robustness across variations in user counts and training sets.
- Highlighted spatial peak shifts as key indicators of AF.
📘 Why It Matters
- Offers non-invasive, interpretable, and high-performing AF detection.
- Bridges signal processing, image transformation, and XAI for clinical diagnostics.
- Suitable for future clinical integration due to simplicity and transparency.