As individuals, it can be hard to look in the mirror, confront our shortcomings, and determine where we can improve. As a society, it’s even more daunting – and it takes the buy-in of many stakeholders at once.

Still, recognizing bias in health care solutions and technology is critical in advancing equitable health innovation. As health care continues to embrace innovation, technology is shaping everything from patient diagnostics to treatment delivery. While these advancements promise better outcomes for all patients, they also contain bias. Embedded in the algorithms and systems we rely on, bias can skew results, perpetuate inequities, and hinder the very progress these technologies aim to achieve.

To improve health care for all, it’s important to understand how bias manifests in health care solutions and technology. Only then can we move to identifying and addressing it. By recognizing and tackling bias, we can work toward a more equitable health care future where innovation truly serves everyone.

The Current Reality of Bias in Health Care Solutions

Health care technology reflects current values and biases in society. Though inadvertently, bias is often ingrained in new technologies because of the datasets used to support them.

A common example of racial bias in health care technology is the pulse oximeter. Pulse oximeters, which are sensors that attach to a finger or toe to measure oxygen saturation in the blood using light, are used everywhere in health care. They support major health care decisions that can mean life or death for patients.

However, this tool is inherently biased – and it has been for decades. It has been shown to overestimate oxygen levels in patients with darker skin tones. Because of this, patient results could read as normal when they are dangerously low.

Published papers documented this bias back in 2005, but it wasn’t until the COVID-19 pandemic that this shortcoming was taken seriously. Even now, there is no clear solution to completely address this glaring inequity in health care.

The pulse oximeter is just one of many examples of bias in health care technology. Artificial intelligence (AI) systems across the country have been found to reflect bias. For example, in one system, Black patients were required to be deemed much sicker than white patients in order for providers to recommend the same type of care. However, the algorithm that was responsible for this was based on data from past healthcare spending – and there is a history of Black people not having as much to spend on care because of income disparities. The algorithm didn’t take actual care needs into account.

Another system was recommending more C-section births, which come with risks for both the mother and baby, on Black/African American and Hispanic/Latine women compared to white women.

These examples are not isolated incidents. To move toward true health equity, it’s critical that health care innovators, researchers, providers, organizations, and patients are aware of and understand how to identify these rampant biases.

Recognizing and Addressing Bias in Health Care Technology

The pulse oximeter is a prime example of how long bias can go unaddressed in health care. It took the racial reckoning of the COVID-19 pandemic to spark real change, and even now, proposed solutions are not perfect.

Short of a global health emergency like COVID-19, we must look to other factors, one being when errors affect specific populations disproportionately. For instance, when more Black/African American and Hispanic/Latine women are getting C-sections, we must ask why and whether it’s necessary.

This requires an enormous amount of time and resources, including teams dedicated solely to reviewing and supervising new technologies through regular audits and interdisciplinary collaboration. To ensure equity remains top of mind, development and review teams themselves should be diverse enough to represent unrepresented populations.

Another sign of bias is limited diversity in clinical data. In order for data to tell a complete story and accurately inform the development of health technologies, it needs to reflect the demographics of the entire population the technology will serve. It isn’t beneficial when a clinical trial only includes subjects from select demographics, such as only white people or only people who are in good health.

However, much of the data that recent technologies have been based on are not representative of all populations – particularly Black people. Black people have been underrepresented in clinical trials, for reasons ranging from not being asked to join them to spiritual beliefs to a mistrust of the medical system that has marginalized them.

By ensuring diversity in clinical trials and including people from historically excluded populations, we can move toward more representative algorithms that accurately reflect the needs of real patients.

On a smaller scale, these responsibilities fall to developers and health care leaders within their own institutions and circles of influence. On a national scale, system-wide solutions include advocating for policies that support equitable technology development. For instance, clinical trials should be incentivized to include diverse populations.

Moving Forward in Health Equity

Bias in health care solutions and technology is not just a technical issue – it’s a human one, with real consequences for patient care and health equity. By identifying where bias exists and actively working to address it, we can ensure innovations that shape the future of medicine truly benefit all people.

Achieving this requires collaboration across industries, commitment to transparency, accountability on all levels, and diversity in every stage of development. By prioritizing inclusivity in health care innovation, we can close gaps in care, improve outcomes, and build a system where every patient gets the care they need.