Despite clear guidelines and effective treatments, diabetic retinopathy (DR) screening remains suboptimal worldwide. In the US, adherence to recommended annual eye examinations can be as low as 20%, with well-documented disparities in access and outcomes across different patient populations. A new study from Johns Hopkins Medicine suggests that autonomous artificial intelligence (AI), deployed at the point of primary care, may help address not only screening rates – but also downstream access to specialist eye care.
The study evaluated 3,745 adults with diabetes referred from primary care clinics to a tertiary eye center via two pathways: either traditional clinician referral or autonomous AI-assisted DR screening. In the AI pathway, patients underwent point-of-care retinal imaging, with referrals triggered by positive or non-diagnostic results. This model effectively embeds DR detection within routine primary care visits, removing the need for separate ophthalmic appointments at the screening stage.
While previous work has shown that such systems improve screening uptake, this analysis focused on a more critical question: do they actually increase attendance at specialist eye care? The answer, at least for certain populations, appears to be yes.
Using propensity score matching to account for demographic and clinical differences, the study authors found that patients screened via AI were more likely to present to an eye care specialist – particularly African-American patients, a group historically underserved in DR care. This is a notable finding, given that African-American patients are less likely to undergo routine screening and more likely to present with advanced disease, including higher rates of diabetic macular edema, vision-threatening retinopathy, and vision loss.
The mechanism is likely multifactorial. Point-of-care AI screening reduces friction in the patient pathway, delivering immediate results and actionable referrals. It may also reinforce the importance of follow-up by making disease risk more tangible to patients at the time of consultation. As illustrated in the study, the AI pathway integrates seamlessly into primary care workflows, triggering referral decisions in real time and potentially improving continuity of care.
The study also highlights a critical nuance involved: improving screening alone is insufficient. Even with perfect screening adherence, the authors note, “outcomes on a population level for DR will still not improve unless we are able to ensure downstream ophthalmic access for further evaluation and treatment for the patients in need.” Autonomous AI, the authors say, can support this transition – at least in part – by increasing the likelihood that at-risk patients engage with ophthalmic services.
However, these findings should still be interpreted with caution. The study is retrospective and exploratory, and conducted within a single integrated healthcare system. As such, causality cannot be established, and external validity may be limited. Nonetheless, the signal is compelling, particularly in the context of ongoing efforts to reduce health inequities in eye care.
The study evaluated 3,745 adults with diabetes referred from primary care clinics to a tertiary eye center via two pathways: either traditional clinician referral or autonomous AI-assisted DR screening. In the AI pathway, patients underwent point-of-care retinal imaging, with referrals triggered by positive or non-diagnostic results. This model effectively embeds DR detection within routine primary care visits, removing the need for separate ophthalmic appointments at the screening stage.
While previous work has shown that such systems improve screening uptake, this analysis focused on a more critical question: do they actually increase attendance at specialist eye care? The answer, at least for certain populations, appears to be yes.
Using propensity score matching to account for demographic and clinical differences, the study authors found that patients screened via AI were more likely to present to an eye care specialist – particularly African-American patients, a group historically underserved in DR care. This is a notable finding, given that African-American patients are less likely to undergo routine screening and more likely to present with advanced disease, including higher rates of diabetic macular edema, vision-threatening retinopathy, and vision loss.
The mechanism is likely multifactorial. Point-of-care AI screening reduces friction in the patient pathway, delivering immediate results and actionable referrals. It may also reinforce the importance of follow-up by making disease risk more tangible to patients at the time of consultation. As illustrated in the study, the AI pathway integrates seamlessly into primary care workflows, triggering referral decisions in real time and potentially improving continuity of care.
The study also highlights a critical nuance involved: improving screening alone is insufficient. Even with perfect screening adherence, the authors note, “outcomes on a population level for DR will still not improve unless we are able to ensure downstream ophthalmic access for further evaluation and treatment for the patients in need.” Autonomous AI, the authors say, can support this transition – at least in part – by increasing the likelihood that at-risk patients engage with ophthalmic services.
However, these findings should still be interpreted with caution. The study is retrospective and exploratory, and conducted within a single integrated healthcare system. As such, causality cannot be established, and external validity may be limited. Nonetheless, the signal is compelling, particularly in the context of ongoing efforts to reduce health inequities in eye care.
