How AI Can Help Alleviate Black Maternal Health Disparities

SDOH DATA NEEDED TO IMPROVE MATERNAL HEALTHCARE IN BLACK COMMUNITIES

One of the major challenges to addressing Black health overall, but especially Black maternal health, is gathering complete high-quality data, including those related to social determinants of health (SDOH). SDOH data, including race and ethnicity, age, and socioeconomic factors, such as income level, education level, access to medical care, prepregnancy health, and a person’s overall health status, play a vital role in maternal mortality disparities, according to Puritz.

“For example, if a woman is low income and doesn’t have access to transportation, she may not be able to obtain early and consistent prenatal care,” Puritz explained. “That’s why early identification is so critical to find the moms early so that we can help them have the healthiest pregnancy.”

“There can be quite a bit of chronic stress present in a woman’s life, such as inadequate housing or even safe housing that can impact her birth outcome because it could contribute to hypertension and preterm labor. So being able to address these at-risk moms before they become pregnant, and at the time of the pregnancy, is key to supporting them so that hopefully some of those stressors can be alleviated,” she continued.

However, relevant data to bolster early identification and patient support are not always available or evaluated thoroughly enough.

“Pervasive disparities exist in the United States healthcare system, resulting from a lengthy history of medical abuse toward marginalized people,” Haywood stated. “As a result, there's a lot of mistrust that healthcare providers need to address in these communities. This includes the SDOH data that healthcare providers can take note of during patient visits.”

“Medical records include geographic data and supplemental ICD-10 SDOH diagnosis codes that document patients’ SDOH. There are codes for everything from houselessness and domestic violence to food insecurity. But healthcare providers don’t always use SDOH screening tools as frequently as they should. Even though it may be uncomfortable to ask patients, the issues they are experiencing can put their overall health at risk,” he continued.

Haywood also noted that the types of data collected are sometimes less important than the depth to which that data is mined. Many healthcare organizations are trying to help pregnant patients as much as possible, but their ability to do so can be limited when they are looking for only surface-level insights or in the wrong place.

HOW AI CAN HELP CLOSE MATERNAL CARE GAPS

If organizations have high-quality, population-based data, equitable algorithms can play a significant role in improving Black maternal health, according to Haywood.

“It’s important not to take general national data and make assumptions and apply it to the local community, but instead to build it from the ground up and make certain they're taken to consideration,” he explained. “This is where a zip code can be just as important as the DNA code. Some communities lack access to grocery stores where pregnant women can buy fruits, vegetables, lean proteins, and other healthy foods. Those are food deserts. Likewise, [areas are] termed maternity deserts if they lack facilities where people can seek maternity care. The best algorithms can be tuned to those specific challenges of a particular geographic area.”

AI algorithms also provide a potential solution for those interested in supporting patients during the 'fourth trimester,' Haywood and Puritz agreed.

“It’s extremely important to collect fourth-trimester data. This is the transition period after childbirth when infants are adjusting to life outside the womb and moms are adjusting to new parenthood. Significant biological, psychosocial, and social changes are occurring during this time, but our current healthcare system doesn’t support it properly,” Puritz stated.

According to Puritz, clinicians can use various interventions to help patients flagged as at-risk by AI, whether it’s a clinical intervention, such as helping the patient understand a medical condition that may be impacting them, or a social one, like helping the patient escape domestic violence and get placed into a shelter.

Health organizations can then use the insights generated by these algorithms to not only measure Black maternal health outcomes but also to determine the success of a maternal health initiative overall.

Using an AI-driven platform, like Lucina’s, allows healthcare organizations to identify a majority of at-risk pregnancies within the first trimester, which improves early intervention efforts, they noted. However, users of any algorithm or platform can measure changes in relevant metrics to look for areas of improvement.

For example, analyzing first-trimester prenatal visit and postpartum uptake, lower preterm birth rates, and decreased incidence of low birth weight newborns and neonatal intensive care unit utilization are how many stakeholders measure performance related to maternal health efforts.

But, at the end of the day, the success of any initiative to address maternal health disparities hinges on the compassion of clinicians.

“Questions about interpersonal safety and other sensitive issues can be uncomfortable, so we have to find ways to make it less so — namely, by removing the associated stigma. Working in patient care, you have to really humanize and empathize with patients to create an atmosphere where they feel safe enough to share with you,” Puritz said.