Your App's Vitals SDK

The same engine that powers NEUROFIT's Vitals feature, optimized for your app. Your user covers the rear camera and flash with a fingertip for 60 seconds, and the SDK returns the metrics and a quality tier.

Validated Against A Chest Strap

HRV (RMSSD) measurements were within 3.4 ms of a Polar H10 chest strap on average, across 30 people and 265 accepted readings. Readings the SDK flagged as unreliable were withheld rather than shown.

Left: camera vs. strap HRV (RMSSD) for every accepted reading. Right: Bland-Altman plot with 95% limits of agreement, dotted lines at ±20 ms.

Consistent Across Skin Tones, Genders And Ages

No group had a reading off by 20 ms or more. Groups with one or two people describe those individuals, not the whole group.

Detects Change Within The Same Person

Each participant took seated readings at rest and during a one-minute slow-breathing exercise at 6 breaths per minute, and we compared each person with themselves.

Averages of each participant's mean, from the SDK and from the Polar H10 strap on the same accepted readings, for the 28 participants with at least two accepted readings in each condition. We expected HRV (RMSSD) and SDNN to rise and the stress index to fall. The SDK and strap changes correlate at r 0.96 for HRV (RMSSD), 0.98 for SDNN, 0.91 for the stress index, 0.93 for SD2:SD1 and 0.99 for heart rate.

Each thin line is one participant (teal iOS, amber Android); the thick line is the average.

The SDK picked up the expected shift: HRV (RMSSD) rose in 27 of 28 participants, by about 18 ms on average, and the stress index fell in all 28, closely matching the chest strap. These are short-term changes, measured seated during the exercise itself in sessions that began with the resting readings, so they don't show lasting effects.

License It Forever

Tell us about your app and the outcomes you want to prove, and we'll walk you through licensing and integration.

Proprietary Approach · Tuned On NEUROFIT Data · Clean Dependencies

How The Study Was Run

Setup

30 volunteers took ten 60-second readings each, half at rest and half breathing at 6 breaths per minute, while wearing a Polar H10 chest strap.

Reference

The strap's beat-to-beat timing. 3 readings with frequent irregular heartbeats were excluded.

What we don't claim yet

Accuracy for Monk skin tones 7, 9 and 10, or during movement. Breathing rate outside paced breathing is an estimate, pending further validation. Most participants were 25–34, and the two investigators are excluded from every figure.

Why A Finger Scan, Not A Face Scan

RMSSD is the HRV number WHOOP, Oura, Polar and Fitbit report, and the hardest one to measure with a camera. It's built from tiny beat-to-beat changes, so any error in timing a beat adds straight to it and makes HRV read high.

That's a physics problem for face scans. On a face, the pulse is a faint color change in reflected room light, easily swamped by shifts in lighting and small head movements. A fingertip pressed over the flash is lit by a bright, steady light right against the skin, giving a pulse about 6× larger, or over 30× the signal power. Our average HRV (RMSSD) error is 3.4 ms, while the best face-video result in a 2023 benchmark was 10.5 ms, with seated, still participants.

Facial pulse: Wang et al., Biomedical Optics Express, 2017 (15 participants, medians read from their Figure 2a), compared with our 30 study participants using the same method. Face-video RMSSD: Bioengineering, 2023 (UBFC-rPPG dataset). Different studies, cameras and references, so not a head-to-head comparison.

Frequently Asked Questions

Interim results, updated as the validation study grows. The NEUROFIT Vitals SDK is not a medical device and does not diagnose, treat or prevent any condition.