AI-assisted eye screening could improve access to care for underserved populations

A new study from the Wilmer Eye Institute at Johns Hopkins Medicine suggests that artificial intelligence may help address long-standing gaps in eye care for people with diabetes. Researchers found that African American patients were more likely to receive referrals for eye exams when screened with an AI-assisted diagnostic tool.

The findings offer early evidence that technology could improve access to care for underserved populations. The results focus on diabetic retinopathy, a leading cause of blindness that often develops without warning.

Diabetic retinopathy is the most common eye disease linked to diabetes. It damages the blood vessels in the retina and can lead to permanent vision loss. Many people do not notice symptoms in the early stages, which makes regular screening critical.

Doctors recommend annual eye exams for people with diabetes. Early detection allows treatment to begin before serious damage occurs. Despite these guidelines, many patients do not receive routine screenings.

Certain groups face higher risks. African American patients and those with limited resources are more likely to develop severe disease. At the same time, they are less likely to receive regular eye exams. This gap has contributed to higher rates of vision loss.

To address this issue, researchers studied an AI-assisted screening program used in community-based primary care clinics. The tool is approved by the U.S. Food and Drug Administration and can analyze retinal images in real time.

During a routine visit, patients have images of their eyes taken with a specialized camera. The AI system evaluates the images immediately. If signs of disease appear, the patient receives a referral to an eye specialist on the same day.

This approach removes a key barrier. Patients no longer need to wait for a separate appointment just to learn if something may be wrong.

“With the AI tool, the patient is evaluated on the spot and given a test result,” said T.Y. Alvin Liu. “They’re not being asked to attend an appointment because they may have something wrong.”

The study examined 3,745 adult patients with diabetes who were referred for eye evaluations between August 2020 and September 2022. Most patients received referrals from primary care providers, while a smaller group was referred through the AI system.

Researchers compared how often different groups received referrals. They also looked at whether patients followed through and attended specialist appointments.

The results showed a clear difference for African American patients. When the AI tool was used, 64.9 percent of these patients received referrals. Without the tool, the rate was 44.4 percent.

This suggests that AI screening increased the likelihood of referral for a group that has historically faced barriers to care.

The study also examined whether referrals translated into real care. Patients who used the AI tool and attended their follow-up visits were more likely to be African American. The increase was about 15 percent compared to traditional referrals.

This finding matters because referrals alone do not guarantee treatment. Patients must attend appointments to receive care. The AI system appears to help move patients through this full process.

Researchers also observed higher referral rates for patients with other health conditions. Those with hypertension and chronic kidney disease were more likely to receive referrals when screened with the AI tool.

These conditions often increase the risk of diabetic eye disease. Identifying these patients early can help prevent complications.

Not all findings showed differences. The study found that Medicaid coverage did not significantly change referral rates or appointment attendance between the two methods.

This suggests that insurance status alone may not determine whether patients receive or follow through with care. Other factors, such as transportation, time, and access to clinics, likely play a role.

The results highlight the complexity of healthcare disparities. While AI may improve some aspects of access, it does not solve every barrier.

One key advantage of AI screening is speed. Traditional referrals often involve delays. Patients may need to schedule another visit, take time off work, or arrange transportation.

The AI system changes that experience. Patients receive results during the same visit. This immediate feedback can influence how seriously they view the need for care.

“But we were able to see that they are more convinced they need care if they’re given immediate results with clear instructions on what to do,” Liu said.

This sense of urgency may explain why more patients follow through after AI screening.

Diabetic retinopathy affects millions of people worldwide. It remains a leading cause of blindness, especially among working-age adults. Early detection can reduce the risk of vision loss by more than 90 percent.

Despite this, screening rates remain low. Many patients miss annual exams due to cost, access, or lack of awareness.

Healthcare systems have searched for ways to close this gap. AI tools offer one possible solution by bringing screening directly into primary care settings.

This approach allows more patients to be evaluated without needing specialist visits. It also helps identify those who need urgent care.

The researchers caution that their findings are exploratory. The study relied on a review of existing medical records rather than a controlled trial. This means it can show associations but not direct cause and effect.

The data also came from a single health system. Results may differ in other regions or populations. Some patients may have received care outside the system, which could affect the findings.

Liu emphasized the need for further research. “Ultimately, AI tools are not meaningful unless you can demonstrate that their real-world deployment positively impacts patient lives.”

Future studies will examine whether improved access leads to better long-term vision outcomes.

The findings point to a broader shift in healthcare. Technology is moving closer to the patient, rather than requiring patients to navigate complex systems.

For underserved communities, this shift could make a meaningful difference. By reducing delays and providing clear results, AI tools may help more people receive the care they need.

The study does not claim to solve all disparities. However, it suggests that targeted use of technology can improve specific gaps in care.

This research highlights how AI can improve access to preventive healthcare, especially for conditions that progress without symptoms. By providing immediate results during routine visits, AI screening can increase patient awareness and urgency. This may lead to higher follow-up rates and earlier treatment.

For healthcare systems, integrating AI tools into primary care could reduce the burden on specialists while expanding screening coverage. This approach may be particularly valuable in underserved areas where access to eye care is limited.

The findings also suggest that technology can help reduce disparities in care. By standardizing screening and referral processes, AI tools may limit the effects of bias or missed diagnoses.

In the long term, broader adoption of AI-assisted screening could lead to fewer cases of preventable blindness. It may also guide future research on how digital tools can improve outcomes across other chronic diseases.

Research findings are available online in the journal npj Digital Medicine.

The original story "AI-assisted eye screening could improve access to care for underserved populations" is published in The Brighter Side of News.

Like these kind of feel good stories? Get The Brighter Side of News' newsletter.