New Study Shows How AI Can Improve Eye Care in Sub-Saharan Africa

Orbis announces new research findings demonstrating that diabetic retinopathy (DR) screenings supported by artificial intelligence (AI) enhanced patients’ speed and uptake of referral services in Rwanda, which can help reduce vision loss due to diabetes in the region. The study is the first globally that explores the use of AI-assisted screening for DR and referral uptake and also includes results that are broadly applicable across other settings for AI-supported DR screenings. The study is published in a special issue of Ophthalmology Science on big data and artificial intelligence in ophthalmic research.

“The number of people living with diabetes is growing rapidly worldwide, with the largest projected increase in Africa at an estimated 143% by 2045. Unfortunately, screening programs for diabetes-related vision loss are often difficult to implement effectively in low-resource settings,” said Dr. Ciku Mathenge, Medical Advisor for Orbis International and the study’s principal investigator.

“Our research findings prove that integrating cutting-edge technology such as artificial intelligence into diabetic retinopathy screenings is not only feasible, but is also an effective solution for identifying those at risk and improving access to care for those with the disease, even in the most vulnerable communities."

The study looked at whether the use of Orbis’s Cybersight AI tool to detect DR in images taken of the eye during routine screenings would lead to patients’ increased uptake of DR referral services. The screenings were conducted in partnership with the Rwandan International Institute of Ophthalmology (RIIO) at four diabetes clinics in and around Kigali, Rwanda, in 2021. Participants, who were recruited during routine visits to the diabetologist, had been diagnosed with type 1 or 2 diabetes and were aged 18 years or older.

The research found that patients who received immediate feedback about their condition based on AI-supported screenings were more likely to go to recommended follow-up appointments than those who received referrals only after a human grading report was completed, a process that can take several days. Potential reasons for this were attributed to a few factors:

• Immediate feedback provides an opportunity for eye care teams to educate a patient about the importance of follow-up care just after a patient learns about their condition.
• Immediate feedback also provides an opportunity for patients requiring referrals to potentially visit the ophthalmologist on the same day, minimizing the need for additional travel for an appointment.
• Receiving an instant report that included images of the retina – a visual reminder of one’s condition – may have also supported acceptance of the recommended referral, especially for patients that are asymptomatic.

The growing burden of diabetes and its associated complications is increasing demands on healthcare systems worldwide. DR, a complication of diabetes that affects the eyes, is caused by damage to the blood vessels to the retina at the back of the eye. Poorly controlled blood sugar is a risk factor. If not treated, diabetic retinopathy can lead to a gradual loss of vision or blindness that cannot be reversed. It is estimated that by 2040, 224 million people globally will have some form of diabetic retinopathy, with vision threatened in 70 million of these individuals worldwide.

Early diagnosis and treatment of diabetic retinopathy through screening reduces vision loss by 98%. But low-resource countries such as Rwanda often do not have sufficient healthcare infrastructure and trained medical personnel to implement diabetic retinopathy screening programs effectively. Furthermore, many patients who are referred for additional follow-up appointments for DR often fail to do so due to financial barriers, travel time, lack of clarity in the referral process and uncertainty about the treatability of the disease.

Recent advances in AI present a promising opportunity for scaling efficient and successful DR programs, particularly in areas where there is a shortage of ophthalmologists as other qualified health care personnel, such as nurses, can also be trained to effectively use the technology. This significantly increases the number of people who can be screened, helps to identify people with diabetes at risk of blindness much faster and reduces unnecessary referrals.

Cybersight, Orbis’s award-winning telemedicine platform, gives eye care professionals in areas with the greatest need free virtual access to training and other resources to better help their patients. One of the platform’s greatest assets is its artificial intelligence tool, Cybersight AI. Cybersight AI can detect abnormalities often associated with common eye diseases – like glaucoma, DR and macular disease – in mere seconds by analyzing images of the back of the eye taken during routine examinations. When diseases like these are caught and treated early, patients have the best chance of not losing their sight.

The newly published research was funded in part thanks to Dr. Mathenge’s receipt of a Roche Collaborative Research Fellowship from the Association for Research and Vision in Ophthalmology (ARVO). The Fellowship pairs early career researchers from developing countries with collaborating scientists in well-established research laboratories with the aim of strengthening ophthalmic research capacities worldwide.