Green-Infrared Dominance in Atrial Fibrillation Detection

Objective: Atrial fibrillation (AF), an asymptomatic arrhythmia, poses a significant risk of stroke when not promptly detected. The primary goal is to analyze the influence of green, red, and infrared wavelengths in the detection and prediction of AF, focusing on the combinations of Green-IR (GIR) and Red-IR (RIR) wavelengths to prevent its occurrence and facilitate timely intervention.Methodology: Using the Multi-Spectral Photoplethysmography (MSPPG) signal obtained from a developed prototype incorporating Red (660 nm), Infrared (IR) (950 nm), and Green (526 nm) wavelengths, AF detection and prediction were conducted using two wavelength combinations: GIR and the commonly available RIR combination, demonstrating improved performance compared to traditional PPG methods. The signals from both combinations underwent feature extraction and classification using the Support Vector Machine with Radial Basis Function (SVM-RBF) classifier. The features extracted at the window size of onset of AF, 5mins before AF, 10mins before AF and 1 hour before the AF event are compared using the statistical analysis of Student’s t-test which demonstrates the capability of differentiating normal and abnormal conditions using both GIR and RIR combinations.Results: The results indicate that the GIR combination consistently outperforms the RIR combination across all window sizes, highlighting the superior efficacy of GIR in the detection and prediction of atrial fibrillation. Notably, the prediction using GIR yielded particularly improved results in the 5-minute window before the onset of atrial fibrillation.Conclusion: The study demonstrates the feasibility of predicting atrial fibrillation, especially within the crucial 5-minute window before onset, using the GIR combination. Early detection within this timeframe offers the potential to prevent atrial fibrillation and subsequent strokes, thereby reducing mortality rates associated with atrial fibrillation and stroke.