Beat-to-Beat HRV
from RR Intervals
Reference for every metric PulseDex computes from a raw beat-to-beat RR-interval series — definitions, formulas, expected ranges, and the evidence base behind each. Because PulseDex ingests genuine RR intervals (not a 1 Hz pulse-rate proxy), its time-, frequency- and Poincaré-domain HRV are the real, literature-validated measures; on top it layers nonlinear/fragmentation signals, readiness composites and population projections. Companion to the analyzer output; not a substitute for clinical evaluation.
Mean heart rate over the recording, computed directly from the RR series.
| Resting HR (bpm) | Read |
|---|---|
| 50–65 | Low–normal (trained/rested) |
| 66–75 | Typical |
| 76–90 | Elevated |
| > 90 | High |
The mean normal-to-normal interval (ms) — the reciprocal basis of mean HR.
Median interval — more robust to occasional outliers than the mean.
SDNN as a percentage of mean RR — HRV scaled to heart rate, useful for cross-session comparison.
Number of accepted intervals after artifact rejection — whether the window is long enough to trust.
| N / duration | Adequacy |
|---|---|
| ≥ 5-min window | Full short-term suite |
| 1–5 min | Ultra-short (rMSSD/HF only) |
| < 1 min | Too short |
Sum of accepted intervals as a fraction of wall-clock span — how much of the window produced beats.
| Coverage | Confidence |
|---|---|
| ≥ 90% | High |
| 70–90% | Moderate |
| < 70% | Low — HRV unreliable |
Share of intervals replaced during cleaning. High artifact attenuates HRV — read every value against it.
| Artifacts | Read |
|---|---|
| < 5% | Excellent |
| 5–15% | Moderate |
| > 15% | HRV unreliable |
The shortest accepted interval — the fastest instantaneous beat after artifact rejection.
The longest accepted interval — the slowest instantaneous beat.
Lower quartile of the interval distribution.
Upper quartile of the interval distribution.
The modal interval — the histogram peak; an input to the Baevsky stress index.
Percentage of beats within ±25 ms of the mode — histogram concentration; higher means more rigid rhythm.
Full spread between the longest and shortest interval (ΔX in Baevsky terms).
Raw count of successive interval pairs differing by more than 50 ms — the numerator of pNN50.
Overall HRV — total autonomic variability over the window.
| SDNN (ms) | Read |
|---|---|
| > 50 | Healthy |
| 30–50 | Moderate |
| < 30 | Reduced |
The primary parasympathetic (vagal) marker — dominated by fast beat-to-beat changes; valid on short windows.
| rMSSD (ms) | Read |
|---|---|
| > 42 | High vagal tone |
| 20–42 | Typical |
| < 20 | Low |
rMSSD on a log scale — the right-skew correction that makes day-to-day readiness changes roughly linear.
Percentage of successive interval pairs differing by more than 50 ms — a vagally-mediated measure.
| pNN50 | Read |
|---|---|
| > 15% | High vagal activity |
| 3–15% | Typical |
| < 3% | Low |
Standard deviation of 5-minute average intervals — slower, circadian-scale variability; needs a long record.
Average of the SDNN computed within each 5-minute segment — short-term variability averaged over a long record.
Total intervals divided by the histogram peak height — a geometric HRV measure robust to occasional artifact.
Integral of the power spectrum — equals the series variance; the spectral counterpart of SDNN².
Power in 0.15–0.40 Hz — respiratory sinus arrhythmia; a parasympathetic marker.
Power in 0.04–0.15 Hz — mixed sympathetic/parasympathetic with baroreflex influence.
Power in 0.003–0.04 Hz — needs a long record to be meaningful.
Ratio of LF to HF power — historically read as sympatho-vagal balance; interpret directionally.
| LF/HF | Read |
|---|---|
| 1–2 | Balanced (typical resting) |
| > 2.5 | Sympathetic-leaning |
| < 0.5 | Parasympathetic-leaning |
HF as a fraction of (LF+HF) — parasympathetic share, less sensitive to total power.
LF as a fraction of (LF+HF).
Width perpendicular to the line of identity — instantaneous beat-to-beat variability, tied to rMSSD.
Spread along the line of identity — longer-term variability, related to SDNN.
Balance of short- to long-term variability.
Area of the fitted Poincaré ellipse — a single geometric summary of overall dispersion.
A geometric stress index from the histogram mode, its amplitude (AMo) and range (MxDMn) — rises with sympathetic load / autonomic rigidity.
| Baevsky SI | Read |
|---|---|
| < 150 | Balanced |
| 150–500 | Elevated load |
| > 500 | High sympathetic load |
Detrended fluctuation analysis over 4–16 beats; α1 ≈ 1 marks healthy correlated dynamics.
| DFA α1 | Read |
|---|---|
| 0.85–1.15 | Healthy correlated dynamics |
| < 0.75 / > 1.25 | Drifting toward random/rigid |
Regularity of the interval series — lower means more repetitive dynamics.
Phase-rectified signal-averaging measure of the heart’s capacity to decelerate — a validated vagal and post-MI mortality marker.
| Decel Cap (ms) | Mortality risk |
|---|---|
| > 4.5 | Low |
| 2.6–4.5 | Intermediate |
| ≤ 2.5 | High (post-MI cohorts) |
The acceleration counterpart of deceleration capacity — a sympathetic-side PRSA measure.
HF power scaled by mean RR — a respiratory-coupling surrogate.
Breathing rate estimated from the HF spectral peak — an RSA-frequency surrogate, not direct airflow.
| Resp (br/min) | Read |
|---|---|
| 10–20 | Typical resting |
| > 20 | Elevated / shallow |
rMSSD scaled by heart rate — a vagal-coupling efficiency proxy.
SDNN as a z-score against the wearer’s own baseline — how today compares to normal for you.
| SDNN Z | Read |
|---|---|
| > 0 | Above your baseline |
| −1 to 0 | Slightly below |
| < −1 | Well below baseline |
Heart-rate fragmentation: share of intervals that are local direction changes — elevated fragmentation associates with AF risk (Costa 2017).
Inverse of the mean run length between inflection points — a fragmentation index.
Share of intervals in short monotonic runs.
Share of intervals in alternating (zig-zag) segments.
A 0–100 readiness headline blending vagal HRV against your recent baseline.
An inverse-HRV stress estimate — rises as variability falls.
A composite energy/activation estimate.
How recovered the autonomic system looks versus baseline.
A recovery index relative to the personal baseline.
A heart-rhythm coherence estimate (spectral peak sharpness).
A composite attentional-readiness estimate.
A sympathetic-branch activation estimate.
A parasympathetic-branch activation estimate.
An LF/HF-derived sympathetic balance ratio.
An HF/LF-derived parasympathetic balance ratio.
A combined sympathetic/parasympathetic balance index.
A weighted readiness blend of energy, focus and coherence.
A composite resilience score combining coherence, vagal HRV and stress.
The gap between stress and focus estimates.
Focus relative to sympathetic activation.
A parasympathetic-efficiency ratio.
A directional overtraining-risk estimate from sustained autonomic load.
A directional autonomic pattern reminiscent of hypertension risk — not a blood-pressure reading.
A recording data-quality score, not a health outcome.
A resting HRV/HR-ratio fitness estimate — a population proxy, not a cardiopulmonary exercise test.
The HR-ratio baseline VO₂ estimate (Uth–Sørensen 2004).
Validation refers to the underlying metric’s validation in published literature, and does not imply validation of the PulseDex implementation against a gold-standard laboratory dataset.
| Metric Category | Status | Basis |
|---|---|---|
| RR statistics, coverage, artifacts | ● Direct measurement | Raw interval statistics after artifact rejection |
| SDNN, rMSSD, pNN50, SDANN, triangular index | ● Literature-based | Task Force 1996 time-domain standards |
| LF/HF/VLF, LF/HF, nu, total power | ● Literature-based | Task Force 1996; Lomb–Scargle for uneven RR |
| SD1/SD2, ellipse area, Baevsky SI | ● Literature-based | Brennan 2001; Baevsky |
| DFA α1, SampEn, decel/accel cap, fragmentation | ◐ Emerging | Peng 1995; Richman 2000; Bauer 2006; Costa 2017 |
| HRV/Stress/Energy/readiness composites | ○ Experimental composite | PulseDex internal blends; no independent validation |
| VO₂max (population proxy) | ○ Population-derived | Population projections / regressions; not measurements |
| Tier | Meaning | Examples |
|---|---|---|
| Core | Headline, universally interpretable | Pulse HR, Coverage, HRV Score, Stress |
| Advanced | Established support, fuller analysis | SDNN, rMSSD, LF/HF, Baevsky SI |
| Research | Exploratory / emerging / composites | DFA, fragmentation, ANS ratios, projections |
| Metric / Formula | Primary Citation | Category |
|---|---|---|
| SDNN, rMSSD, pNN50, SDANN, triangular index, LF/HF | Task Force 1996 | HRV |
| SD1/SD2 Poincaré geometry | Brennan 2001 | HRV |
| Baevsky stress index | Baevsky & Chernikova 2017 | HRV |
| DFA α1 | Peng 1995 | Nonlinear |
| Sample entropy | Richman & Moorman 2000 | Nonlinear |
| Deceleration / acceleration capacity | Bauer 2006 | Nonlinear |
| PIP / IALS / PSS / PAS fragmentation | Costa 2017 | Fragmentation |
| Lomb–Scargle spectral estimation | Lomb 1976 / Scargle 1982 | Spectral |
| VO₂max (HR-ratio) | Uth–Sørensen 2004 | Projection |
| HRV/Stress/readiness; ANS-age & BP proxies | PulseDex internal — no external source | Internal |
| Formula | Source / Author | Year | Reference |
|---|---|---|---|
| Time-domain HRV standards | Task Force ESC/NASPE | 1996 | Circulation. 93(5):1043–65 |
| SD1/SD2 = f(SDNN, ΔNN) | Brennan, Palaniswami & Kamen | 2001 | IEEE TBME. 48(11):1342–7 |
| Baevsky SI = AMo/(2·Mo·MxDMn) | Baevsky & Chernikova | 2017 | Cardiometry. 10:66–76 |
| DFA α1 | Peng et al. | 1995 | Chaos. 5(1):82–7 |
| SampEn = −ln(A/B) | Richman & Moorman | 2000 | Am J Physiol. 278(6):H2039–49 |
| Deceleration capacity (PRSA) | Bauer et al. | 2006 | Lancet. 367(9523):1674–81 |
| Fragmentation (PIP/IALS/PSS/PAS) | Costa, Davis & Goldberger | 2017 | Front Physiol. 8:255 |
| VO₂max = 15·(HRmax/HRrest) | Uth, Sørensen et al. | 2004 | Eur J Appl Physiol. 91(1):111–5 |
- HRV depends on posture, breathing rate and time of day
- Ectopic beats and artifact must be corrected before HRV is valid
- Frequency-domain & SDANN need adequate window length
- Compare only matched protocols (e.g. same morning routine)
- Absolute HRV varies widely with age and fitness — track trends
- LF/HF as “balance” is contested; read directionally
- Fragmentation/DFA are device- and length-dependent
- No EEG — no sleep-stage claims
- Readiness/stress/energy scores are internal composites
- VO₂max is a population proxies, not measurements
- Normative bands are general, not personally calibrated
- Heavy correction attenuates HRV — check Artifacts %
- Not FDA cleared or CE marked as a medical device
- Not for clinical diagnosis or treatment decisions
- Personal, research, and wellness use only
- Discuss medical concerns with a qualified clinician
| Method / Metric | Primary Citation | Category |
|---|---|---|
| SDNN, rMSSD, pNN50, SDANN, triangular index, LF/HF bands | Task Force of the ESC and NASPE. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation. 1996;93(5):1043–65. doi: 10.1161/01.CIR.93.5.1043 | HRV |
| Poincaré SD1 / SD2 geometry | Brennan M, Palaniswami M, Kamen P. Do existing measures of Poincaré plot geometry reflect nonlinear features of HRV? IEEE Trans Biomed Eng. 2001;48(11):1342–7. doi: 10.1109/10.959330 | HRV |
| Baevsky stress index | Baevsky RM, Chernikova AG. Heart rate variability analysis: physiological foundations and main methods. Cardiometry. 2017;10:66–76. doi: 10.12710/cardiometry.2017.10.6676 | HRV |
| DFA α1 | Peng C-K, Havlin S, Stanley HE, Goldberger AL. Quantification of scaling exponents… in nonstationary heartbeat time series. Chaos. 1995;5(1):82–7. doi: 10.1063/1.166141 | Nonlinear |
| Sample Entropy | Richman JS, Moorman JR. Physiological time-series analysis using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol. 2000;278(6):H2039–49. doi: 10.1152/ajpheart.2000.278.6.H2039 | Nonlinear |
| Deceleration / acceleration capacity (PRSA) | Bauer A, Kantelhardt JW, Barthel P, et al. Deceleration capacity of heart rate as a predictor of mortality after myocardial infarction. Lancet. 2006;367(9523):1674–81. doi: 10.1016/S0140-6736(06)68735-7 | Nonlinear |
| Heart-rate fragmentation (PIP / IALS / PSS / PAS) | Costa MD, Davis RB, Goldberger AL. Heart rate fragmentation: a new approach to the analysis of cardiac interbeat interval dynamics. Front Physiol. 2017;8:255. doi: 10.3389/fphys.2017.00255 | Fragmentation |
| Lomb–Scargle periodogram (uneven RR) | Lomb NR. Least-squares frequency analysis of unequally spaced data. Astrophys Space Sci. 1976;39(2):447–62. doi: 10.1007/BF00648343 | Spectral |
| VO₂max (HR-ratio method) | Uth N, Sørensen H, Overgaard K, Pedersen PK. Estimation of VO₂max from the ratio between HRmax and HRrest. Eur J Appl Physiol. 2004;91(1):111–5. doi: 10.1007/s00421-003-0988-y | Projection |
| HRV / Stress / Energy / readiness scores; BP & ANS-age proxies | PulseDex internal composites and population projections — no external source. Directional only. | Internal |
Implementation · Validation · UI/UX
Literature synthesis · Reference formatting
Planicka M. PulseDex: Beat-to-Beat RR HRV Node. Version 1.0.0. 2026.