FOUR-CHANNEL PPG WEB APP · STEVENS ET AL. 2024
UNO R4 disconnectedUNO R4 WiFi · A0–A3 · 5V · 500 Hz

Four-channel direct Web Serial acquisition

Heart rhythm in the foreground.
Respiration only when the evidence earns it.

Heart rateNeuroKit-style
--BPM
10 s moving meanWAIT
RespirationCharlton fusion
--BrPM
Needs ≥ 20 sWAIT
Signal gateSelected browser channel
0/ 12ACQUIRING
Span --No clipping
Raw transportPPG4 v1 + CRC-16
0.0 host Hz0.0 device Hz0.0 frames/s0 frames
0 timepoints · 0.00 s0 gaps · 0 order · 0 timing · 0 backtrack · 0 malformed · 0 CRC

CH1 / waiting for device

Selected live photoplethysmogram

RawCleanedDetected peaks
−10 s−5 snow

Running estimates

Rate trend

60 s view · local only

3 heart methods × 5 respiratory methods

The paper’s comparison matrix, running in JavaScript.

Reconstruction boundaryThe paper did not publish code or exact parameters. Names below indicate method families, not package-equivalent output.

Live method matrix

Respiratory estimate by heart peak source

Waiting for signal

Recreate the measurement sequence

Five minutes to settle. Three minutes per posture.

READY

05:00Acclimation

Sit comfortably, keep the dominant wrist relaxed, and let the signal stabilize before measurement.

    Published study cohort

    15 healthy volunteers

    Age
    30.26 ± 5.23 y
    Gender
    6 M · 9 F
    Height
    174.33 ± 8.8 cm
    Wrist
    17.6 ± 0.9 cm
    Positions
    4 × 3 minutes

    The original research used four PulseSensors on the dominant wrist plus a Vivalink ECG patch. This build acquires four independent PulseSensor channels through the Arduino UNO R4 WiFi reference sender.

    Sensors 24(12):3766 · DOI 10.3390/s24123766

    What the authors found, without sanding off the caveats.

    Guylian Stevens, Luc Hantson, Michiel Larmuseau, Jan R. Heerman, Vincent Siau, and Pascal Verdonck. “A Guide to Measuring Heart and Respiratory Rates Based on Off-the-Shelf Photoplethysmographic Hardware and Open-Source Software.” Sensors 24(12):3766 (2024).

    Open paper ↗
    Best seated HRρ 0.90

    NeuroKit · mean difference 0.59 BPM

    Best seated RRρ 0.82

    NeuroKit + Charlton · mean difference 1.90 BrPM

    Walkingρ 0.29 / 0.31

    Best reported HR / RR correlations remained low

    Protocol15 × 12 min

    Supine · seated · standing · walking in place

    Tables 3–6

    Published Bland–Altman and Spearman results

    MeasureHeart methodRespiratory methodMean diff.95% intervalρ

    Hardware + interpretation boundary

    A transparent instrument for research and education.

    Supported acquisition device

    Arduino UNO R4 WiFi

    Signals 1–4A0 · A1 · A2 · A3
    Shared power5V
    Shared groundGND

    The reference sender samples all four analog inputs on one 500 Hz timer, then sends CRC-protected PPG4 frames through direct Web Serial. 12-bit · 500 Hz · PPG4 v1

    PulseSensor Lab standard: UNO R4 WiFi.Power every PulseSensor from 5V and share one GND.

    Original paper hardware

    Four green-light reflection sensors

    • Sensors4 × PulseSensor
    • ControllerArduino Nano 33 IoT
    • InputsA0–A3
    • PlacementNarrow line, wrist underside
    • ReferenceVivalink Cardiac Patch
    • TransportBLE to a personal computer
    Why the authors chose PulseSensorTable 1 compared four named alternatives. PulseSensor was the only listed green-light board, provided raw data, and was selected as the optimal choice based on wavelength and price.

    Known limits

    What this build can and cannot claim

    • ✓The implementation supports display, recording, and export of four aligned 500 Hz channels; connected CSV export remains an acceptance gate.
    • ✓Runs transparent approximations of all 15 HR/RR combinations on the selected channel.
    • ✓Recreates the posture timing and one-minute summary workflow.
    • —No exact replication of unpublished Python parameters or participant data.
    • —No automatic cross-sensor fusion or claim that one sensor placement is best.
    • —The UNO R4 connected transport/display path passed with operator-declared electrically connected A0/A1 and unconnected/floating A2/A3; four-sensor isolation and physiological validation remain pending.

    Source trail

    What informed this dashboard

    1. Stevens et al. 2024Guylian Stevens, Luc Hantson, Michiel Larmuseau, Jan R. Heerman, Vincent Siau, and Pascal Verdonck. “A Guide to Measuring Heart and Respiratory Rates Based on Off-the-Shelf Photoplethysmographic Hardware and Open-Source Software.” Sensors 24(12):3766 (2024). DOI 10.3390/s24123766.
    2. Arduino UNO R4 WiFi senderOriginal timer-driven A0–A3 acquisition and compact PPG4 transport built on ArduinoCore-renesas APIs.
    3. PulseSensor communityPulseSensor hardware ecosystem and multi-sensor precedent; no library beat detector is compiled into this project.
    4. Browser reconstructionOriginal readable JavaScript where paper parameters were unavailable; cited package/gist source is not copied.

    Open the full provenance and method-lineage map. The paper’s data availability statement says participant data are available on request from the corresponding author.