At a primary care clinic in Baltimore, a patient's routine diabetes visit takes an unexpected turn toward prevention. A specialized camera captures images of the back of their eye, an AI tool analyzes the results in real time, and if diabetic retinopathy is detected, they walk out the door with a referral to an eye specialist—not a vague suggestion to "schedule an appointment someday," but an actionable path forward with an appointment offered that same day. This small shift in workflow, tested at Johns Hopkins Medicine's Wilmer Eye Institute, has revealed something significant: an FDA-approved AI diagnostic tool is helping close a critical health gap for African American adults with diabetes.
The disparity is stark and urgent. Diabetic retinopathy—the most common eye disease linked to diabetes—causes blindness globally and strikes African Americans and other racial minorities at disproportionately high rates. Yet these same communities are paradoxically less likely to receive the annual eye exams that catch the disease early, when treatment can still preserve sight. A referral from a primary care doctor doesn't always translate into action. Patients may delay, forget, or struggle to navigate the health system. The question researchers asked was whether AI could change the equation.
Led by Dr. T.Y. Alvin Liu, principal investigator and founding director of the James P. Gills Jr., M.D., & Heather Gills Artificial Intelligence Innovation Center at Wilmer, the team analyzed data from 3,745 adult patients with diabetes who came in for diabetic retinopathy evaluation between August 2020 and September 2022. Of these, 3,352 received traditional referrals from their primary care doctors, while 393 received referrals after AI screening at the point of care. The findings, published in npj Digital Medicine on April 13, showed a striking difference for African American patients: 64.9% received an eye exam referral when the AI tool was used, compared to just 44.4% through standard referrals alone—a 46% increase.
The mechanism matters as much as the outcome. When patients opted for AI screening, retinal images were captured with a specialized camera and analyzed during their primary care visit. Real-time feedback meant no waiting, no uncertainty about whether they "might" need to see a specialist. They got their answer and their referral on the spot. This immediate, concrete action appears to have altered the calculus for patients, particularly African Americans already navigating systemic barriers to eye care.
The results extended beyond African American patients. Those with hypertension were more likely to receive referrals with the AI tool (89.6% vs. 82.6%), as were those with chronic kidney disease (26.2% vs. 20.9%). Dr. Liu emphasized the psychological dimension: "With the AI tool, the patient is evaluated on the spot and given a test result. They're not being asked to attend an appointment because they may have something wrong." That reframing—from suspicion to diagnosis—appears to empower patients to act.
The findings don't solve all barriers to eye care. Insurance and transportation obstacles remain real. Yet they suggest a practical pathway for how artificial intelligence, thoughtfully deployed, can help address entrenched health inequities. By removing friction at a critical moment of decision-making, an AI tool became a bridge to care for patients who needed it most.