A 20-minute cuff-and-sensor test reads heart risk differently in men and women

A blood-pressure cuff on the upper arm inflates for a few minutes, briefly cutting off the flow of blood to the forearm. A small optical sensor resting on the skin beneath it watches the tiniest blood vessels underneath, the microvasculature, as the cuff releases. In a new [microvascular sub-study of 3,177 Swedish adults, published in Biology of Sex Differences](https://pubmed.ncbi.nlm.nih.gov/42509570/), the optical signal that separated lower-risk women from higher-risk women was not the same signal that worked in men.

That inversion is the cleanest current evidence that cardiovascular risk in women has been chronically underdetected, not because women's hearts are different in some essential way, but because the canonical tools were largely developed on male-majority cohorts.

The microvascular sub-study sits inside SCAPIS, the Swedish CArdioPulmonary bioImage Study. The sample is 3,177 people: 1,645 at low cardiovascular risk (no diabetes, no prior atherosclerotic disease, and a SCORE2 score below 5%) and 761 at high risk (SCORE2 above 7.5%, or diabetes, or prior atherosclerotic disease, or severe chronic kidney disease). SCORE2 is the European risk calculator that estimates a person's ten-year risk of a heart attack or stroke.

The test itself is non-invasive. A cuff on the upper arm is inflated for a few minutes and then released, briefly stopping and restoring blood flow. During that 20-minute protocol, two complementary optical techniques watch the tissue underneath the sensor. Laser Doppler flowmetry measures how much blood is moving through the tiny vessels, what researchers call perfusion. Diffuse reflectance spectroscopy measures how saturated that blood is with oxygen, the oxygen saturation. Together, they record a time-resolved picture of the post-occlusion hyperemic response: the rush of blood and oxygen back into the tissue when the cuff releases.

The researchers fed those time-resolved signals into a random forest classifier, a machine-learning pattern finder that can be inspected to see which features did the work. They trained one model on the full cohort and two separate models by sex.

In the full cohort, the model reached an AUC of 0.685 (95% CI 0.663–0.707). An AUC of 1.0 is a perfect discriminator; 0.5 is a coin flip. 0.685 is meaningfully better than chance, but well short of what a clinician would call a screening-grade test.

The sex-stratified numbers were closer together than the story they tell. In women, the AUC was 0.638 (95% CI 0.596–0.679). In men, 0.644 (0.628–0.660). The discrimination is modest in both. The split underneath is not.

In men, both perfusion and oxygen saturation contributed to the model's discrimination, with low-risk men showing a more pronounced hyperemic perfusion response, a stronger blood-flow rebound after the cuff released. In women, the discrimination came almost entirely from oxygen saturation.

The signal that separates lower-risk women from higher-risk women is not the signal that works in men. That is the structural finding.

It is consistent with a long-standing concern. The two most widely used heart-risk calculators in clinical practice, SCORE2 in Europe and the ASCVD (atherosclerotic cardiovascular disease) risk score in the United States, were derived from cohorts that were majority male. When those tools are applied to women, they tend to underestimate risk, particularly in younger and middle-aged women, where heart disease has historically been under-recognized. The microvascular finding does not fix that. But it suggests the gap is not simply a matter of where the threshold is set; the underlying signal that carries the risk information may be different.

The cohort is a single Northern-European population (SCAPIS, in Sweden). The discrimination is modest, well below screening-grade, and the protocol remains research-stage, not a clinical test. The paper does not claim otherwise.

A sex-aware reading of the microvascular signal is possible: the answer to what heart risk looks like under the skin depends on whose skin is being read. The next milestone is whether the female-driven oxygen-saturation pattern holds in a second cohort outside Scandinavia.