The grey oximeter on my kitchen table looks like a turntable. As a product of the 1970s, its low blood oxygen level alarm is set via an analog dial. I bought it on eBay at the end of last year, after writing a story for the Boston Review about racial bias in blood oxygen saturation measurement. Meir Kryger, a professor of medicine at Yale University, then made a suggestion: I should also study models that are older than contemporary pulse oximeters, especially those made by HP. It is a technical dinosaur, but in some respects, its internal working principle is more advanced than many of the devices that measure blood oxygen in today’s hospitals.
For decades, researchers have proven that many pulse oximeters commonly used in hospitals do not meet the FDA’s safety thresholds for people of color. Because these devices use optical color sensing to assess oxygen in the blood, they may make mistakes for people with darker skin due to racial prejudice during the calibration process. However, when Covid-19 first appeared, pulse oximeter readings were still hailed as a “biomarker” during early hospitalization and triage. Some people of color who told emergency room doctors that they had difficulty breathing were actually sent home when the device indicated that they did not need oxygen.
It wasn’t until a team of University of Michigan doctors reinvestigated the device in December that the wider medical community began to pay more attention. “When the pulse oximeter shows 91% [oxygen saturation], more than 50% of black patients actually have values lower than 88%,” said study co-author Tom Valley. Since then, the senator and the FDA began to discuss this issue, which aroused strong interest from doctors and engineers as well as confused patients. There is now a widespread call for a redesign of the “pulse cow”—and rethinking the review system that has failed to detect or prevent these errors for decades.
However, these key debates about the possible future models often still overlook an important fact: oximeters are designed to work fairly regardless of skin color, gender, and disability, and they actually existed as early as the 1970s. However, somehow, most of the history of the equipment in my kitchen has been erased.
As early as the 1960s and 1970s, HP worked closely with NASA engineers to create equipment for the health of astronauts, in which the accurate measurement of oxygen played a vital role. When the company expanded into the hospital market, it carefully designed an oximeter that could similarly be calibrated for individual patients. It is based on a transmission method (to pass light through tissue), which combines optical fibers and eight wavelengths of light. (Many current pulse oximeters use only two wavelengths.) Hewlett-Packard engineers knew that the device needed different brightness settings to work consistently across a range of skin tones. The company provided a transparent and thoughtful discussion — still available today in its communications archive — on how it built a light-sensing technology that works equally well for all skin tones.
As a step to alleviate racial prejudice, HP engineers have assembled a series of more inclusive oximetry methods. The baseline calibration of the instrument was set in collaboration with a “carefully selected” team, which included 248 black volunteers-notably, this is more than the FDA’s current recommendations for pre-market testing of oximeters in hospitals 246 non-whites. Most importantly, the device can be adjusted for everyone. You can choose to squeeze a small drop of blood from the wearer’s ear to scan the blood using spectrophotometry. This measurement helps to accurately identify how much light is absorbed by an individual’s skin and tissues, allowing doctors to personalize the light level calibration and optimize the accuracy of the device.
The oximeter can also explain circulatory characteristics. Unlike modern pulse cows, which only test healthy people, HP’s equipment is designed to work for people who may be sick. For example, the sensor is not designed for fingertips, because in this way, the device will not work properly for patients suffering from common health conditions such as shock, sepsis and certain chronic diseases. Instead, HP placed its sensors on the top curve of the ear, which is one of the last parts of the body affected by circulatory problems during illness. This choice helps prevent the incorporation of capacityism into oxygen measures, while also avoiding gender differences due to improper equipment. Although ear oximeters still exist in the professional world, by far the most common models in emergency rooms and homes today are not adjustable and designed to fit the “average” geometry of male fingers, sometimes for all others Produce sub-optimal readings, which may be related to other errors.
Despite these achievements, when the personal computing market exploded in the 1980s, HP shifted its focus and withdrew from medical equipment shortly before releasing the long-planned miniature version of the oximeter. But Kryger still describes its larger device as “the best oximeter ever.” Publications in his laboratory at the time indicated that HP oximeters were more accurate in many respects than the pulse oximeters that soon replaced them. They are called the non-invasive “gold standard” for testing early pulse oximetry in clinical studies because HP oximeter readings are closer to invasive arterial blood gas testing.
As the pandemic reminds us painfully, the consequences of this inaccuracy can be devastating. Since today’s hospital oximeters do not have personalized functions, they may inadvertently provide defective data to doctors and other machines. The oximeter numbers provide key inputs to a range of computing systems, including algorithms that guide ICU classification and certain insurance reimbursements. They also use closed-loop algorithms with many ventilators-such devices may not be optimized effectively when inputting errors-ridden inputs.
These conversations are now essential: As an increasingly important part of artificial intelligence in healthcare, various non-invasive sensors based on pulse oximeters are being developed. Some, such as some optical sensors for sepsis or blood sugar, may already be in your local hospital or at your home. If you are not careful, the next-generation optical color sensor can easily reproduce the unequal errors in pulse oximetry measurements that are now known in many other medical fields.
We tend to assume that technology will unfold in a linear way, and useful features or key issues will be built into future models. The history of equipment is often written later, as if it has always been the case-alternative methods have not been successful because they are inferior. But like any history, it’s useful to ask who wrote it and what was missing.
You can tell about these missing oximeter functions as an accidental case: When HP shifted its focus from medical equipment in the 1980s, most small companies that entered the field of hospital equipment did not have years of cooperation with NASA. A multidisciplinary experience that is widely used. Therefore, when American companies started adding the “pulse” of Japanese bioengineer Takuo Aoyagi to the oximetry model, they adopted his insights instead of conducting public-facing accounting as their predecessors did for the US market. Many hospitals that purchased these devices for the first time did not even realize that the newer oximeter lacked the functions that existed in other models before, because the “pulse cow” was their first exposure to non-invasive oxygen measurement.
This loss of technology can also be told as a story about changing history and social norms over time. HP’s model in the 1970s included a transparent discussion of fair design—but it was also carried out in the context of the difficult results of the civil rights movement, when the issue of racial justice was being discussed more openly across departments. At the same time, the later impulse cow model (the first in the United States to be patented by Biox in 1980) became another face of the corporate enclosure of that era. Kryger recalled that when they first entered the market, he tried to ask engineers for calibration data, which was once considered a standard safety practice. But they will not share it again. “They are black boxes hidden by proprietary algorithms,” Kryger said. “The engineers at the time would not provide any technical information, such as accuracy for people with dark skin pigments.”
People usually call today the era of precision medicine. But the HP oximeter proved a more disturbing and unbalanced story.
When black boxing began to further invade medical technology in the 1980s and 1990s, doctors conducted a series of studies on the accuracy of pulse cattle. However, in the end, doctors got used to not knowing and stopped asking certain questions about the biases built into the device. The blind spot is getting bigger and bigger. Princeton University professor Ruha Benjamin (Ruha Benjamin), while studying the inclusion of racial bias in the scope of hospital algorithms, commented in the journal Science: “Today, coding inequality persists precisely because of design and adoption. People with such tools did not carefully consider systemic racism.” People usually refer to today as the era of precision medicine. But the customizable HP counter-model proves a more disturbing and unbalanced story of what happens when data and design decisions are put into a black box, despite potential errors, but separated from public responsibility.
No matter how you imagine it, re-learning this history is an opportunity to imagine a different future. In addition to this pandemic and other pandemics that may occur, oximeters are used every day at critical and delicate moments, such as childbirth, which have become notorious for magnifying racial inequality. During Covid, Pulse ox’s long-standing design issues became more difficult to ignore, which reminded people of the continuing risks of retrofitting it—and the root cause of its bias. Today, health policy expert Onyeka Otugo and the recent multinational intensive care interns who wrote in The Lancet, a new generation of doctors, together with Kadija Ferryman and Mikaela Pitcan, and other multidisciplinary engineers and health sociologists, have raised questions about hospital technology Again, a new insight into the design policy.
Some people, such as Kryger, still remember what they wrote from later blood oxygen saturation measurement history: it is possible to design color sensing devices with a basic commitment to fair design. For medical safety and fairness considerations, companies can be required to publicly share complete calibration and accuracy data. It is possible to design the most suitable hospital equipment for all patients. HP’s oximeter completed all these things nearly half a century ago. The one sitting on my kitchen table doesn’t have all the proper retro cables, so it doesn’t really work. But this became an important reminder for me: fair device design has been achieved before. It needs to happen again.
WIRED is where tomorrow is realized. It is an important source of information and ideas, making the world meaningful in constant change. Wired conversations shed light on how technology can change every aspect of our lives-from culture to business, science to design. The breakthroughs and innovations we discovered brought new ways of thinking, new connections and new industries.
© 2021 Condé Nast. all rights reserved. By using this website, you accept our user agreement and privacy policy and cookie statement, as well as your California privacy rights. As part of our affiliate partnership with retailers, Cable may obtain a portion of sales from products purchased through our website. Without the prior written permission of Condé Nast, the materials on this website may not be copied, distributed, transmitted, cached or otherwise used. Ad selection
Post time: Jun-22-2021