INTEGRATING EEG AND HEART RATE VARIABILITY PATTERNS TO PREDICT ATRIAL FIBRILLATION IN STROKE PATIENTS: AN AUTONOMIC TASK-BASED APPROACH

Embolic Stroke of Undetermined Source (ESUS) accounts for approximately 20% of all ischemic strokes, with previously undetected atrial fibrillation (AF) assumed to be a major underlying mechanism. AF frequently occurs in paroxysmal and asymptomatic forms, making detection difficult with standard diagnostic procedures. Therefore, early identification of individuals at elevated risk for AF is essential to prevent recurrent strokes. Evidence increasingly supports brain–heart interactions using integrated neural and cardiac signals may assist AF risk stratification.
This study aimed to develop predictive markers for AF risk by examining electroencephalographic (EEG) features, and heart rate variability (HRV) parameters patterns during autonomic challenge tasks designed to modulate cardiac activity. The sample included 19 ESUS patients (mean age 69.6 ± 6.2 years) and 35 healthy elderly controls (mean age 63.1 ± 7.0 years).
Our analysis revealed robust time-dependent modulation of heart rate (HR) across all experimental conditions, indicating effective autonomic engagement with a significant Group × Time interaction observed during the Isometric Exercise task. HRV measures showed significant group differences in Pre vs. Post Cycling, where indices such as RMSSD, and Sample Entropy (SampEn) demonstrated significant Group × Time interactions, reflecting altered autonomic recovery dynamics. EEG spectral analyses revealed significant main effects of group and interactions with time and region, particularly in frontal, temporal, centroparietal and occipital regions in theta-alpha and beta frequencies. Overall, these findings indicate task-dependent differences in autonomic and neural dynamics between groups, which provide preliminary evidence that multimodal brain–heart signatures may enhance AF risk detection and support further investigation in larger clinical cohorts.