Accepted for/Published in: JMIR Cardio
Date Submitted: May 26, 2025
Open Peer Review Period: May 26, 2025 - Jul 21, 2025
Date Accepted: Sep 4, 2025
(closed for review but you can still tweet)
Photoplethysmography-Based Machine Learning Approaches for Atrial Fibrillation burden：Pilot study (Pre-mAFA IV Registry)
ABSTRACT
Background:
Atrial fibrillation (AF) burden is associated with cardiovascular events such as stroke and heart failure. Recent advancements in photoplethysmography(PPG) technology have provided new insights into non-invasive and convenient AF burden detection.
Objective:
This study aims to establish an AF burden model based on smartwatch-monitored PPG technology to track the progression of AF in real-world settings.
Methods:
This prospective study (January 2024 to January 2025) at Chinese PLA General Hospital enrolled patients with paroxysmal AF. Participants underwent simultaneous rhythm monitoring using smartwatch PPG and 24-hour Holter ECG (the gold standard).Five PPG-derived AF burden metrics were defined:M1: AF episode duration/total monitoring time.M2: AF episode frequency / total measurements .M3: AF episode density.M4: AF episode Variability.M5: Proportion of rapid ventricular rate AF episodes (>120 bpm).Smartwatch PPG signals were collected once per minute. Sensitivity, specificity, and accuracy were used to evaluate the PPG algorithm's AF detection capability by comparison with the gold standard (24-hour Holter monitoring). Mean absolute error (MAE) and correlation coefficients (r-values) were used to assess the correlation between the PPG-based AF burden metrics and the gold standard.
Results:
A total of 145 participants with paroxysmal AF (66% male, mean age: 63.28±14.23 years) were included. Compared to the gold standard, the PPG-based AF burden model demonstrated a sensitivity of 91.5% (95% CI :87.9%-95.1%), specificity of 97.2% (95% CI :95.9%-98.5%), and accuracy of 95.7% (95% CI:94.0%-97.4%).M1: MAE for the model of AF episode duration as a proportion of total monitoring time was 0.04, with a correlation coefficient (Rs) of 0.8788 (P< 0.001).M2: MAE for the model of AF episode frequency as a proportion of total measurements was 0.032, with a correlation coefficient (Rs) of -0.0807 (P= 0.334).M3: MAE for the AF density model was 0.1725, with a correlation coefficient (Rs) of 0.6576 (P< 0.001).M4: MAE for the AF episode variability model was 3.9967, with a correlation coefficient (Rs) of 0.7876 (P< 0.001). MAE for the average real variability model was 4.6436, with a correlation coefficient (Rs) of 0.8127 (P< 0.001). MAE for the average AF change model was 0.3893, with a correlation coefficient (Rs) of 0.7246 (P< 0.001).M5: MAE for the model of AF episodes >120 bpm as a proportion of total monitoring sessions was 0.0151, with a correlation coefficient (Rs) of 0.3435 (P< 0.001). MAE for the model of AF episode duration >120 bpm as a proportion of total monitoring time was 0.0151, with a correlation coefficient (Rs) of 0.3435 (P< 0.001).
Conclusions:
The PPG-based AF burden model demonstrates high concordance with the gold standard of 24-hour Holter monitoring in tracking AF episode duration and variability, providing new perspectives for exploring AF progression dynamics. Clinical Trial: Chinese Clinical Trial Registry ChiCTR2300075516
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.