Scientists work to correct skin color bias in pulse oximeters.
During the height of the COVID-19 pandemic, the news that pulse oximeters were consistently overestimating the blood oxygen levels of Black patients—and thus delaying their care—made headlines.
However, it wasn’t exactly news. Experts had known for decades that skin pigmentation and melanin can affect a pulse oximeter’s ability to accurately measure oxygen saturation, an indicator of pulmonary health. It just wasn’t widely known, and as a result, inaccurate readings likely contributed to the disproportionate mortality of patients with darker skin during the pandemic.
Kimani Toussaint, PhD, the Thomas J. Watson, Sr., Professor of Science, is leading an effort to improve this important clinical screening tool so that patients and physicians can trust the reading.
“The focus is always the same thing: trying to develop a more equitable device,” says Toussaint, the senior associate dean for research and strategic initiatives in Brown’s School of Engineering.
The pulse oximeter exhibits biases for two main reasons. First, FDA guidelines for calibrating the devices require at least 15 percent of participants to identify as “darkskinned”—but they don’t specify how skin tone should be measured, or define “dark skin tone.” Second, melanin not only absorbs ambient light but also contributes to light scattering, complicating measurement.
“The organelles produced by melanosomes in the epidermis vary in size, density, and distribution depending on skin tone,” Toussaint says. “This means that, in addition to absorption, melanin-induced scatter also affects the measurement.” Together, these factors can cause the pulse oximeter to overestimate oxygen levels.
Toussaint’s lab is trying to exploit the polarization properties of light so that devices are less sensitive to the absorption and scattering phenomena at the top epidermal layers at the measurement site, thus emphasizing contributions from the deeper layers of the skin.
The latest generation of the instrument, which is roughly the size of a shoebox, is undergoing clinical trials at The Miriam Hospital in an outpatient setting. Toussaint’s team is also translating their design into a small-scale wearable that goes around the wrist. Rutendo Jakachira ScM’22 PhD’26, a physics student in his lab, says that as light travels through their device’s larger optical system, the power decreases.
“By the time the light is incident on the finger, we have lost a lot of power. This adds another layer of complexity when dealing with patients with low perfusion,” Jakachira says.
Working on a smaller scale allows for a small footprint between the light source and detectors, resulting in a more prominent and accurate signal—but it’s more costly, as they are “essentially building a whole new device as we iterate,” she says.
Toussaint says the inaccurate pulse oximeter is only “the tip of the iceberg” of health disparities in device manufacturing.
“It’s a blind spot that is indicative of not having sufficient diversity across the entire development of this type of technology,” Toussaint says. “And it’s all rolled up into this health equity piece that technology itself can have bias irrespective of the original intent.”
