We don't ship a vital sign until we've benchmarked it against a reference sensor on hundreds of subjects across skin tones, ages, genders, and lighting conditions. Here's the evidence.
Mean absolute error (MAE) across our validation cohort (N = 1,247), compared to FDA-cleared reference sensors.
Scatter of SmartScanPro rPPG HR vs. 3-lead ECG HR across 1,247 paired 60-second observations. Line of unity overlaid. r = 0.97.
J. Biomed. Informatics, 2025 — A prospective study of 612 primary-care patients showed rPPG HR non-inferior to finger-pulse-oximetry across skin tones (p = 0.31).
Read PDF2024 — We open-sourced 840 labeled 60-second face-scan clips paired with ECG, SpO₂, and blood-pressure references. Apache-2.0. Use it to compare any rPPG system.
DownloadarXiv 2024.xxxxx — Fairness audit across Fitzpatrick I–VI with a learned melanin-aware colour model. Max subgroup MAE gap 0.8 bpm.
ReadIEEE EMBC, 2024 — 298 post-op subjects. Our respiratory rate within ±1.5 rpm of capnography 93 % of the time.
ProceedingsAge, gender, Fitzpatrick skin type, BMI, indoor/outdoor lighting — stratified sampling per metric.
FDA-cleared or clinical-grade reference devices recording simultaneously for the full scan window.
Bland-Altman limits, Pearson r, MAE, RMSE. Per-subgroup breakdowns. Published raw data.
Email research@smartscanpro.ai — we also provide a HIPAA-compliant research sandbox.