Sample Entropy for PPG Complexity Analysis
Sample Entropy (SampEn) measures the conditional probability that two sequences of m consecutive data points that are similar (within tolerance r) will remain similar when one more point is added. Applied to PPG IBI series, SampEn quantifies cardiac rhythm complexity — lower values indicate more regular, predictable rhythms associated with disease.
SampEn is computed as SampEn(m, r, N) = -ln(A/B) where A is the number of template matches of length m+1 and B is the number of matches of length m, with match tolerance r (typically 0.2 × SD of the data). Parameters m = 2 and r = 0.2×SD are standard for HRV analysis per PhysioNet recommendations. SampEn improves upon Approximate Entropy (ApEn) by eliminating self-matches and reducing bias for short time series.
In healthy cardiac rhythms, SampEn values of 1.0–2.0 indicate complex, information-rich dynamics arising from multiple interacting regulatory mechanisms. Disease states reduce complexity: heart failure (SampEn 0.5–0.8), post-MI (0.6–1.0), and severe autonomic neuropathy (0.3–0.7). During AF, SampEn paradoxically increases (1.5–2.5) due to the random irregularity of fibrillatory conduction, distinguishing it from the reduced complexity seen in other cardiac pathologies.
PPG-derived SampEn shows strong concordance with ECG-derived SampEn at rest (ICC = 0.90–0.94). The metric is relatively robust to occasional IBI detection errors because the tolerance parameter r accommodates small perturbations. Multiscale entropy (MSE), which computes SampEn at multiple time scales, provides richer characterization of physiological complexity loss in disease and aging.
Frequently Asked Questions
How does sample entropy differ from approximate entropy?
Sample entropy (SampEn) eliminates the self-match counting bias in approximate entropy (ApEn), producing lower and more consistent values for short time series. SampEn has become the standard entropy measure for HRV analysis, replacing ApEn in most modern studies.
What parameters should be used for PPG SampEn?
m = 2 (embedding dimension) and r = 0.2 × SD of the IBI series are standard. N ≥ 200 data points (beats) are recommended for reliable estimates. These parameters are consistent across the HRV literature and enable cross-study comparisons.
Can SampEn detect AF from PPG?
SampEn increases during AF (>1.5) due to random irregularity, while it decreases in other arrhythmias with regular patterns. Combined with RMSSD and coefficient of variation, SampEn improves AF detection specificity by distinguishing random AF irregularity from other irregular rhythms.