How to address medical AI bias to innovate health equity for all

Addressing medical AI bias: ©Wladimir1804 - stock.adobe.com

Systemic racial biases persistently shape the health care experiences of Black, Latinx, and Indigenous populations. These biases are especially pronounced for women of color, influencing everything from diagnosis and family planning to treatment recommendations. In fact, CDC data shows that Black women are two to three times more likely to die from pregnancy-related complications than non-Hispanic white women, with most of the maternal deaths being preventable and harming the population regardless of income and education level. These alarming trends are present at the earliest stages of life, with Black infants being four times as likely than non-Hispanic white infants to die from low birth weight complications. When addressing health care disparities, it is important to acknowledge the role that representation plays in the space. Whether it is decreasing Black newborn mortality, improving cardiovascular health, or simply building trust to further adhere to medical guidelines and create a safe place for patients of color, there is power in representation and what it can do for historically overlooked communities.

When it comes to the implementation of new artificial intelligence tools in the health care space, oversight is necessary, and representation must be required before the products reach consumers. This must be done to ensure that the biases entrenched in our current systems aren’t inadvertently bolstered or perpetuated by such technologies. This past year, a study from the Stanford School of Medicine revealed that AI chatbots were producing racist health information with little to no oversight, which can further perpetuate disparities and worsen health outcomes for individuals of color. Therefore, at every stage, from funding to reaching the lives of consumers, new health care AI tools must be overseen for racial and gender biases.

Within the asset allocation space, there is an opportunity to catalyze change. Investments supporting fund managers and founders who are overlooked and underrepresented are often perceived as risky due to the limited trust that is historically given to underrepresented individuals in positions of power. To this day, less than 2% of all venture capital goes towards people of color and women in asset management. Due to several hurdles of bias, there is limited trust in having representation in asset management, causing an even wider gap for founders of color who are more likely to consider their community when creating their tech tools. Within the medical space, we are asking patients from diverse communities to trust AI tools with their lives and health outcomes. These are the AI tools that are often built and backed by groups that did little to consider communities of color as customers in the creation of their products. In order to create equitable health care processes using AI, we must place ourselves in the shoes of the communities that are often overlooked and underrepresented to understand how to build this trust and to further build tools with diverse communities in mind. Ultimately, we cannot expect AI to become an equitable health care tool without addressing the deeply ingrained racial and gender biases that continually cost lives and damage health outcomes.

The overall health care venture capital market is vast, having raised $14B in the first half of 2023, potentially bringing it near 2021’s record pace. When looking for solutions that target these biases in the medical space, it is important to question fund managers who are sourcing and funding these AI and tech solutions. It is necessary to understand whether fund managers are prioritizing race and gender disparities in their decisions and how capital is being moved to solutions that are improving patient outcomes. We must consider how fund managers are making their decisions and what kind of questions they are asking in their diligence process before they allocate millions of dollars into the next big AI solution. By addressing biases, there is an opportunity to unlock impact and create economic value through driving patient outcomes.

Within the health care space, leaders must question a product before it reaches their patients and ask how their products are benefiting the communities they aim to engage with. We have already seen the positive effects of having representative health care providers. The next step to implementing equitable health care tools is to prioritize the diverse experiences of communities which shape the input for all AI tools, advance the technological tools, and improve patient outcomes. We must create solutions patients can trust, and AI tools are no different. A future of truly equal health care involves eradicating the biases that plague our current health system and creating tools that improve patients’ lives. Technology has the opportunity to expand access, reduce costs, and refine therapeutics in numerous ways that can address new markets and diverse consumers – but it must be used with the intention of equity. Furthermore, an inclusive and optimal health care industry is the cornerstone of an equitable future.

Daryn Dodson is the managing director of Illumen Capital.

Opinions expressed are those of the writer and do not reflect the official position of Illumen Capital.