PolarH10Windows-first telemetry toolkit
PolarH10 Developer Reference

HRV Formula Sheet

Use this sheet when you want RMSSD, SDNN, pNN50, SD1, mean NN, mean HR, and lnRMSSD written out directly with the repo's short-term window assumptions.

Download Markdown Download PDF

Data Path

  • input stream: accepted RR intervals from live BLE heart-rate measurement packets
  • physical H10: RR intervals are derived inside the strap from ECG beat timing
  • synthetic transport: RR remains the source for the HRV solve, while synthetic PMD ECG is emitted on the same synthetic beat schedule for operator review

Rolling Short-Term Window

The app uses a rolling RR window with a default length of 300 s. That default tracks the conventional short-term setting discussed by Shaffer and Ginsberg.

The first solve is intentionally delayed until the rolling window is nearly full enough to avoid publishing an early headline value from a partial capture.

Time-Domain Formulas

For accepted RR intervals NN_1 ... NN_n:

Here NN_i means the i-th accepted beat interval in milliseconds, n is the number of accepted intervals in the current rolling window, and an overline means a window average. These formulas are applied to the accepted short-term series the tracker keeps for the HRV solve, not to every raw packet value.

NN‾=1n∑i=1nNNi\overline{NN} = \frac{1}{n}\sum_{i=1}^{n} NN_i

This is the average accepted beat interval over the current short-term window.

HR‾=60000NN‾\overline{HR} = \frac{60000}{\overline{NN}}

This converts the mean interval into beats per minute using 60000 ms = 1 min.

SDNN=1n−1∑i=1n(NNi−NN‾)2SDNN = \sqrt{\frac{1}{n-1}\sum_{i=1}^{n}(NN_i - \overline{NN})^2}

SDNN measures how widely the accepted intervals are spread across the whole window.

RMSSD=1n−1∑i=2n(NNi−NNi−1)2RMSSD = \sqrt{\frac{1}{n-1}\sum_{i=2}^{n}(NN_i - NN_{i-1})^2}

RMSSD focuses on beat-to-beat change by differencing neighboring intervals before averaging.

ln⁡RMSSD=ln⁡(RMSSD)\ln RMSSD = \ln(RMSSD)

\ln RMSSD is the natural-log form of RMSSD; it compresses the scale and is often easier to compare across sessions.

pNN50=100×#{∣NNi−NNi−1∣>50 ms}n−1pNN50 = 100 \times \frac{\#\left\{|NN_i - NN_{i-1}| > 50\,ms\right\}}{n-1}

pNN50 is the percentage of adjacent interval pairs whose absolute difference exceeds 50 ms.

SD1=RMSSD2SD1 = \frac{RMSSD}{\sqrt{2}}

SD1 is the short-axis Poincare equivalent of short-term variability and is shown because many readers know HRV through that geometric view.

Alignment Status

The implementation is aligned to the cited short-term HRV guidance for:

  • RMSSD as the headline short-term vagally weighted metric
  • SDNN as sample standard deviation over the accepted short-term window
  • pNN50 with a fixed 50 ms threshold
  • SD1 = RMSSD / sqrt(2)
  • mean NN, mean HR, and lnRMSSD

The app does not claim to reproduce:

  • 24 h norms
  • clinical artifact adjudication
  • full Holter-style NN editing

It is therefore documented as short-term RR-derived telemetry rather than a clinical HRV workstation.

References