Scientists have developed an AI-assisted tool that analyzes blood proteins to predict early retinal nerve damage in people with diabetes. The model, called Pro-DRN, could shift diabetic eye care from reactive treatment toward early prevention.
For the more than half a billion people living with diabetes worldwide, one of the most feared complications is the gradual loss of vision — often arriving silently, long before a patient notices anything is wrong. Now, a new study suggests that signals hidden in a routine blood draw could give doctors a crucial head start in identifying who is at risk.
Published June 2 in the open-access journal PLOS Medicine, the research describes an AI-powered predictive model built on 71 distinct blood plasma proteins that are associated with diabetic retinal neurodegeneration (DRN) — the breakdown of nerve tissue in the retina caused by diabetes. The model, named Pro-DRN, outperformed the previous best predictive approach by 26%, and its creators have already made it available online for physicians to use.
Why Retinal Nerve Damage Is a Bigger Problem Than It Sounds
The retina is the light-sensing layer at the back of the eye, and in people with diabetes it can deteriorate over time in ways that go far beyond blurry vision. Researchers describe DRN as a kind of early warning signal — or “window” — into broader nervous system damage elsewhere in the body, including cognitive decline, dementia and nerve damage in the hands and feet.
The critical challenge is timing. By the time DRN is currently detectable, the damage is already irreversible. That gap between biological change and clinical detection is exactly what the research team, led by Huangdong Li from the Guangdong Provincial Clinical Research Center for Ocular Diseases in Guangzhou, China, set out to close.
How the Study Was Conducted
The team drew blood plasma samples from 1,492 patients enrolled in the Guangzhou Diabetic Eye Study — all diagnosed with type 2 diabetes and showing no signs of DRN at the outset. Researchers monitored the retinal health of 1,218 of those participants through eye scans over a six-year period. They then validated their findings against a separate group of 502 people with diabetes drawn from the UK Biobank.
From that data, they pinpointed 71 plasma proteins linked to DRN development. Many of these proteins are involved in cell pathways tied to inflammation and cellular maintenance — biological processes already known to play a role in diabetic complications. Using machine learning, the team translated those protein-level patterns into Pro-DRN, a risk-stratification tool that doctors can apply using a standard blood test run through an AI system.
What the Authors Say
“Our study suggests that early retinal nerve damage in diabetes leaves measurable signals in the blood. By combining plasma proteomics, longitudinal retinal imaging, and explainable AI, Pro-DRN may help move diabetic eye care from detecting established damage toward earlier, molecularly informed risk stratification, so that closer monitoring and future neuroprotective interventions can be directed to the people most likely to benefit,” the authors said in a news release.
The authors are careful to note that Pro-DRN is based on statistical associations between protein levels and DRN rather than proven cause-and-effect relationships. Still, the team believes the model could pave the way for earlier interventions — and potentially, for neuroprotective treatments not yet in widespread use.
Why This Matters for Students and Young Adults
Type 2 diabetes is increasingly being diagnosed in younger people, including college-age adults, driven by sedentary lifestyles, dietary patterns and rising obesity rates. For students managing or at risk of developing the condition, tools like Pro-DRN represent a meaningful shift in how preventive care could work — moving from annual eye exams that catch damage too late to a blood-based screening that flags risk years in advance.
More broadly, the study is a strong example of how AI and proteomics — the large-scale study of proteins — are beginning to reshape medicine. Rather than waiting for symptoms, clinicians may soon be able to identify biological signatures of disease long before patients feel anything at all. For a generation that has grown up hearing about personalized medicine, this kind of research is a tangible step toward that reality.
The study was conducted by researchers in China, the United States and Australia, and was supported by the National Natural Science Foundation of China and several other public research funding bodies.
Source: PLOS