Our Validation Results: Non-Invasive Screening That Holds Up

We said we would do the study first and report the numbers after. The study is done, and here is what it found.

Between October and December 2025, we ran a cross-sectional validation study at Anubhav Life Care, our NABL-accredited centre in Barasat. We screened 182 adults and 175 completed every measurement — both the SamaClip screen and the reference-standard tests it was compared against. Participants were recruited across Monk Skin Tone categories 4 to 10, including the darkest tones, from the community the centre already serves.

The headline: the painless screen agreed closely with hospital-grade instruments, and it held up across skin tones.

What the numbers show

A few results stand out, all measured against accredited reference instruments.

For anaemia, the non-invasive haemoglobin estimate correlated with lab haemoglobin at r=0.91, and as an anaemia screen reached an area-under-curve of 0.92. In plainer terms, it detected anaemia with roughly 89% accuracy — strong for a needle-free first-line screen in a population where, among the women in our cohort, 51% were anaemic.

For oxygen saturation, the screen showed a mean bias of just −0.31% against a clinical-grade oximeter, and the agreement held within tight bounds across every skin-tone group, including the darkest. This is the result we cared about most, because it is exactly where standard oximeters have historically failed. Accurate across all the skin tones we serve was the bar, and the data cleared it.

For blood-sugar risk, the HbA1c proxy correlated with the lab method at r=0.87 — enough to flag people who should get a confirmatory test. For heart rate, agreement with a full ECG was near-perfect at r=0.98. And for heart rhythm, the AFib classifier reached an AUC of 0.94, catching dangerous irregular rhythms with high accuracy in the arrhythmia subset.

The unglamorous numbers matter too: the vast majority of recordings reached acceptable signal quality on the first try, which is what makes the device usable in a real camp rather than only on a bench.

What this means, and what it doesn't

It means SamaClip is a reliable first-line screen for anaemia, oxygen saturation, blood-sugar risk, heart rate, and rhythm — painless, smartphone-powered, and validated on the population it is built for, including the skin tones the field usually leaves out.

It does not mean the screen is a diagnosis. These results support using SamaClip to flag who needs a closer look. Confirmation still comes from accredited testing and clinical judgement, and a flagged result routes into exactly that. We will keep saying this because it is the line that keeps the work honest.

These figures belong to SamaClip, the finger sensor that was studied. We are not extending them to any other device or feature. A different tool earns its own evidence on its own study.

The full validation manuscript, with the per-skin-tone breakdown, the limits of agreement, the completion rates, and the study's limitations, is available for anyone who wants the detail behind the summary.

FAQ

How accurate is SamaClip at detecting anaemia? In our validation study, the non-invasive haemoglobin estimate correlated with lab haemoglobin at r=0.91 and detected anaemia with an AUC of 0.92 (about 89% accuracy), measured against an accredited reference instrument.

Does it work on dark skin? Yes. Oxygen-saturation accuracy held within tight bounds across Monk Skin Tone categories 4 to 10, including the darkest tones — the range where standard oximeters have historically underperformed.

Is SamaClip a diagnosis? No. It is a screening tool that flags who needs confirmatory testing. Diagnosis comes from accredited lab testing and clinical assessment.