AFib Detection via RR Irregularity Analysis
PPG-based atrial fibrillation detection exploits the hallmark irregularly irregular ventricular rhythm of AF by analyzing statistical properties of consecutive interbeat intervals — coefficient of variation, RMSSD, sample entropy, and Poincaré plot dispersion — to distinguish AF from normal sinus rhythm with 90–98% accuracy.
AF produces chaotic atrial electrical activity that conducts irregularly through the AV node, creating unpredictable ventricular intervals. This translates to highly variable pulse-to-pulse intervals in PPG with characteristic statistical signatures: elevated coefficient of variation (CV > 0.10 vs. < 0.05 in NSR), high RMSSD relative to mean IBI, increased sample entropy (loss of regular periodicity), and dispersed Poincaré plots with SD1 >> SD2 (cloud pattern vs. ellipse).
Classical rule-based AF detection applies thresholds to these features: CV > 0.10 AND RMSSD/meanIBI > 0.06 AND absence of organized periodicity in autocorrelation achieves sensitivity 90–95% and specificity 88–93% on controlled datasets. Machine learning classifiers using 20+ IBI-derived features improve specificity to 95–98% by learning to distinguish AF from other irregular rhythms (frequent PACs/PVCs, sinus arrhythmia). Apple Watch's Irregular Rhythm Notification uses a proprietary classifier analyzing 65 minutes of tachogram data with 84% PPV in the Apple Heart Study.
The key challenge is distinguishing paroxysmal AF (intermittent episodes) from frequent ectopic beats, which produce similar IBI irregularity. Morphological PPG features (loss of consistent dicrotic notch in AF, variable pulse amplitude) combined with rhythm features improve specificity for AF vs. ectopy classification, achieving AUC 0.96–0.99 in deep learning models processing raw PPG waveforms.
Frequently Asked Questions
What is the minimum AF episode duration detectable by PPG?
Most validated algorithms require 30–120 seconds of continuous AF for reliable detection. Brief paroxysmal episodes (<30 seconds) are frequently missed. Continuous monitoring increases detection probability by capturing longer AF episodes.
Can PPG differentiate AF from frequent premature beats?
Frequency of premature beats (PACs/PVCs) confounds simple irregularity metrics. Advanced algorithms using both rhythm statistics and waveform morphology distinguish AF from frequent ectopy with 85–92% specificity, but clinical ECG confirmation remains necessary.
What is the false positive rate for PPG AF screening?
In general population screening, PPG AF algorithms produce 5–15% false positive notifications. The Apple Heart Study showed 16% false positive rate (notifications without confirmed AF on follow-up ECG patch). False positives are higher in younger populations with physiological sinus arrhythmia.