Short-Term Arrhythmia Prediction Using AI Based on Daily Data From Implantable Devices: Multicenter Prospective...

Area of Science:

• Cardiology
• Artificial Intelligence
• Predictive Medicine

Background:

• Predictive medicine utilizes algorithms for personalized treatment strategies.
• Artificial intelligence (AI) shows promise in identifying atrial fibrillation (AF) episodes.
• Current AI models rarely focus on short-term dynamic prediction of arrhythmias.

Purpose of the Study:

• To evaluate an AI model for predicting short-term onset or worsening of arrhythmias.
• To assess the utility of remote monitoring data from pacemakers for arrhythmia prediction.

Main Methods:

• A multicenter prospective observational study involving 314 patients.
• Analysis of 65,243 data sequences from pacemaker remote monitoring.
• Training an AI model on 31-day records to predict arrhythmia changes over the subsequent 14 days.

Main Results:

• The AI model achieved a global sensitivity of 66.4% and specificity of 77.4%.
• For patients with baseline arrhythmia, sensitivity was 76.8% and specificity was 39.6%.
• For patients without baseline arrhythmia, sensitivity was 39% and specificity was 81%.

Conclusions:

• The AI model can predict short-term arrhythmia episode fluctuations using remote monitoring data.
• The model demonstrated reasonable sensitivity and specificity for arrhythmia prediction.
• Future improvements are expected with larger datasets including demographic and clinical information.