BeatAI: BiomEtrics for Atrial Arrhythmia Tracking Using Artificial Intelligence
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Postoperative atrial fibrillation (POAF) affects 20 to 50% of patients undergoing cardiac surgery and is associated with longer hospital stays and adverse outcomes. Although several risk factors for developing POAF have been identified, accurate prediction remains challenging. Wearable ECG patches and remote patient monitoring enable continuous heart rhythm surveillance. Using AI models, subtle yet distinct patterns may be recognized that precede POAF development.
Objective
This study evaluates whether combining continuous ECG patch monitoring with deep learning algorithms can improve both early risk stratification and near real-time prediction of POAF.
Methods
We analyzed continuous ECG and wearable-derived physiology from 20 postoperative cardiac surgery patients enrolled in a prospective monitoring trial. Each patient wore a 14-day adhesive patch sensor (VivaLNK VV-330) capturing per-second ECG and activity streams. Two complementary deep learning pipelines were developed: (1) a daily-level multimodal Transformer, which downsampled ECG and contextual “TAB tokens” into day-wise units to predict AF occurrence and burden, and (2) an hour-ahead forecasting model, which condensed the last two hours of minute-level physiology into attention-weighted summaries to generate rolling, causal predictions of AF risk in the next hour.
Results
Across 162,217 downsampled data elements, the daily-level model showed conservative behavior with very low false negatives, consistently identifying AF-positive days and correctly stratifying high-burden episodes. The hour-ahead forecasting model was trained on 5,607 windows and achieved excellent discrimination (AUC 0.95), high specificity (0.98), and strong predictive value (NPV 0.98). Recall-focused calibration further reduced missed AF hours while maintaining low false alarm rates. Together, the two frameworks provided reliable daily burden stratification and fine-grained, near real-time risk forecasts.
Conclusion
Continuous multimodal monitoring combined with AI enables accurate POAF detection, daily risk stratification, and rolling hour-ahead forecasts. This dual-resolution framework has the potential to support perioperative decision-making by enabling earlier intervention, more precise surveillance, and better allocation of preventive therapies in cardiac surgery patients.
