Wearable stress biofeedback — from wristband firmware to app
A stress measurement and guided relaxation platform: a wrist-worn sensor tracks stress live while a guided session plays in the room.
The challenge
Reflective wrist PPG is a hostile signal. Perfusion on this band measured 0.02–0.25%, and the pulse was routinely weaker than the baseline wander it competed with, so datasheet defaults were never going to produce a heart rate anyone could trust.
Our approach
Five layers, built end to end: ESP32-C3 firmware driving an optical pulse sensor; a signal chain of ambient subtraction, band-pass filtering, adaptive noise cancellation, FFT peak picking and Kalman tracking; RMSSD-based stress scoring; an Electron kiosk runtime in the room; and a React Native app and Next.js admin dashboard on a Node backend. Every constant was set from captures on real wrists against hand-counted pulses.
The results
At rest the band reads 89.7 BPM against a hand-counted 90 and holds lock through typing. Vigorous motion still defeats it — the pulse genuinely disappears — and closing that gap needs accelerometer-based cancellation rather than more filtering.
Locum staffing marketplace — clinicians verified at the source
Let's build something that lasts.
Tell us what you're working on. We'll come back within two business days with a clear next step — not a sales script.