AI is riddled by bias, especially in healthcare. Just one well-known example is a study from 2019 that revealed racial bias in a clinical algorithm used by hospitals showing that Black patients had to be significantly sicker than white patients to receive the same care. The bias stemmed from training data reflecting historical healthcare spending disparities between Black and white patients.
From racism to ageism, these biases have already had profound implications in health outcomes, particularly for older adults. This is not an issue of the future but an issue of today. With biased healthcare algorithms being used every day, impacting who receives medical care and support, algorithmic bias is a public health issue that needs to be addressed with urgency as health tech continues to push forward.
One untapped resource resides within our communities. By designing community-centric programs that bridge public health and healthcare, we have an opportunity to address the inequities AI in healthcare continues to perpetuate.
Build on innovative community approaches
Community-centric programs that reflect the local health priorities must be a part of AI advancement in healthcare, and early models are already emerging. One such program is the Coalition to End Racism in Clinical Algorithms (CERCA), led by Michelle Morse, M.D., and the New York City Department of Public Health, which works to support and accelerate efforts to eradicate explicit and implicit racism in clinical algorithms.
Recognizing that it’s not just individual patients who are negatively impacted by this bias, but rather entire communities, CERCA is addressing racial bias as a part of New York City’s efforts to treat racism as a public health crisis. The very structure of the organization has required community-level collaboration across clinical experts, operational and technical support staff, data analysts and public health communication experts.
This level of multidisciplinary and multi-institutional coordination is often the key ingredient missing in how we approach care; it’s also the program’s primary strength. While the program was launched in 2021 and is ongoing, CERCA has catalyzed change across seven health systems since its inception, leading to the adoption of algorithmic adjustments that promote racial equity.
The early promise of this program underscores the importance and effectiveness of such initiatives in creating tools for oversight and the equitable use of algorithms in healthcare. It offers an opportunity for imitation not only from region to region, but also in addressing other types of bias as well, such as ageism, especially for those in marginalized communities.
Ageism in AI remains pervasive
While there have been advances in acknowledging and starting to address gender, race and other types of bias, data that includes the experiences and information on older adults remains systematically excluded from data collection and data sets relied upon in AI development and other data-driven technology innovation.
Research suggests that bias is most frequently introduced during the data ingestion phase of machine learning and when assessing representation and evaluation processes of algorithm development. As a result, these innovations and solutions either neglect older adults or generate inaccurate predictions about them.
With approximately 1 in 6 Americans aged 65 and over as of 2020 and projections indicating a doubling of this demographic by 2040, the need for age-inclusive healthcare innovation has never been more pressing.
Collaborate to build a sustainable framework
For a more equitable, impactful healthtech ecosystem, programs like CERCA should not be the exception but the norm. To get there, standardized systems and processes, including playbooks like the White House’s U.S. Playbook to Address Social Determinants of Health (SDOH) (PDF) and the Algorithmic Bias Playbook created by Berkeley Public Health Professor Ziad Obermeyer, M.D., and colleagues at the University of Chicago Booth School of Business will be critical for advancement. Further, organizations such as the Coalition for Health AI and the Trustworthy & Responsible AI Network are underscoring the growing efforts to ensure algorithms and AI are being developed fairly and safely with equity in mind.
Collaborative efforts such as these must involve community organizations, public health departments, tech industry leaders and government entities. This confluence is essential to create a universal framework for equitable healthcare AI inclusive of the needs of older adults, especially those in marginalized communities.
Transform rhetoric into action
Amid the rhetoric surrounding health equity and the buzz of AI, we need more programs and initiatives to provide a concrete pathway for action such as the ones highlighted earlier. By dismantling biased algorithms and fostering inclusivity in healthcare innovation, we can ensure that advancements in healthcare technology benefit all segments of society, including the lived experiences of older adults.
It’s time to translate intentions into impactful change, tapping into our communities and the important work of engaged organizations across the U.S. in order to drive health equity forward, one algorithm at a time.
Anika Heavener is vice president of innovation and investments at The SCAN Foundation, a philanthropic organization working to ensure all Americans can age well with purpose.