QRS Detection & HRV Analyzer

Inspect an offline QRS processing pipeline, match peaks to synthetic truth or MIT-BIH annotations, and quantify time-domain HRV.

Chapter 7
QRS
HRV
MIT-BIH validation

Sensitivity

100.0%

Positive predictive value

100.0%

TP / FP / FN

34 / 0 / 0

Reference / detected

34 / 34

QRS processing and detected beats

raw

Time-domain HRV — reference versus detected

Metric	Reference NN (synthetic truth / annotated)	Detected RR (may include errors)
Mean RR (ms)	856.73	857.95
SDNN (ms)	24.53	27.34
RMSSD (ms), contiguous NN only	30.01	34.49
pNN50 (%)	0.00	3.13

Detected tachogram

Detected RR

Poincaré plot of detected RR

Model scope

Offline Pan–Tompkins-inspired detector, 5–18 Hz FFT bandpass, 150 ms causal integration, adaptive threshold, 200 ms refractory period after peak snapping. Original causal filters, search-back, and T-wave discrimination are not implemented. Zero-padding/FFT boundary effects remain. Thirty seconds is insufficient for reliable frequency-domain HRV; no LF/HF or clinical diagnosis is reported.

Preparing the simulation…

Conceptual bridges & related modules

12-Lead ECG & Vectorcardiography
companion · Chapter 7
Distinguish waveform-generation assumptions from beat-detection and interval errors.
Bioelectric Uncertainty, Stimulation & EIT
companion · Chapter 7
Separate interval repeatability, fiducial definition, and sampling uncertainty.
Classification Metrics Explorer
downstream classifier · Chapter 9 · Section 7
Understand feature-extraction errors before using cardiac features in classification.
Cost-Sensitive Threshold Explorer
downstream classifier · Chapter 9 · Section 7
Compare error costs in detection and downstream decision thresholds.