Orbis International releases new research supporting artificial intelligence (AI) detection of diabetic retinopathy

The technology proves to be a practical solution in low-resource communities for tackling the leading cause of vision loss among working-age adults. (Image Credit: ©ipopba - Adobe.Stock.com)

Orbis International is taking the wraps off new research supporting the use of artificial intelligence (AI) to detect diabetic retinopathy in patients.

According to an Orbis news release, the study¹ was conducted in Rwanda in partnership with the Rwanda International Institute of Ophthalmology, and the research demonstrates that AI-based screening for diabetic retinopathy is not only feasible at diabetes clinics in sub-Saharan Africa, but also leads to high levels of patient satisfaction.

Published in the British Journal of Ophthalmology, this is one of the first clinical studies on the practical applications of AI in diabetic retinopathy screening in Africa.¹

"This study has the potential to make a big impact for patients with diabetes in Africa," Ciku Mathenge, PhD, medical Advisor for Orbis International and the study's principal investigator, said in the news release. "In the study, we found that most patients had not had a recent full eye exam despite worrying about their vision and knowing their risks for developing diabetic retinopathy. Our research findings show that access to easy-to-use and conveniently located screening technologies allows patients to get the care they need and to prevent vision loss or blindness from diabetic retinopathy. The key now will be to ensure that such technologies become affordable and accessible to all."

According to Orbis, the research examined the satisfaction levels of patients screened for diabetic retinopathy and other eye conditions using Orbis's Cybersight AI tool. The screenings were conducted during routine appointments at four diabetes clinics in and near Kigali, Rwanda, in 2021.

Moreover, Orbis noted in its news release the research found that patient satisfaction with AI screening was high, at over 99%. More than 63% of the study participants preferred AI over human graders. Factors that may have contributed to high participant satisfaction include:

• Receiving the exam during their diabetes appointment, as opposed to scheduling a separate appointment with an eye clinic, saved time and the cost of travel, an important barrier for rural patients, who composed nearly half the participants.
• The AI screening model was integrated into the existing workflow at the four diabetes clinics where the study took place.
• Most patients were not required to have their pupils dilated to undergo AI screening, which reduces time commitment and avoids temporarily blurred vision.
• Printed reports, available immediately during the screening, provided an opportunity for patients to learn about the condition and their risks while at their appointment, as opposed to waiting for delayed reporting from human grading.
• The study showed Cybersight AI performed well and led to accurate referrals from diabetes clinics.

Orbis also noted that diabetes is on the rise globally, and diabetic retinopathy is the leading cause of vision loss among working-age people around the world. Recent reports show the prevalence of diabetic retinopathy is increasing in sub-Saharan Africa, more than in any other region. The organization noted that access to diabetic retinopathy screening is key to preventing visual impairment as most patients with diabetes do not realize they have the disease until their vision is affected irreversibly. Low-resource settings, such as Rwanda, lack the infrastructure and trained personnel to implement the screenings effectively.

According to the news release, sub-Saharan Africa has a lower number of ophthalmologists, averaging 3.7 physicians per 1 million people. Training other medical personnel to conduct diabetic retinopathy screenings is an option and effective model in low- and middle-income countries that preserve ophthalmologists' time for care only they can manage. AI makes the screening process accessible for medical personnel and patients alike in non-ophthalmic settings.

"It was like a miracle," said Etienne Uwingabire, a senior nurse and Director of the Rwanda Diabetes Association, after receiving the AI screening camera at the clinic in Kigali. "This camera gives a direct response, and the patient is very happy to know the status of their eyes."

Orbis noted in its news release Uwingabire and the diabetes clinic team spent two sessions learning the ins and outs of the camera. Uwingabire can now supervise other members of the clinic in the use of the technology.

"Anyone can do it [AI screening] with the training," Raban Susabimana, MD, a physician at the Rwanda Diabetes Association clinic who recently completed his residency and was trained to use the AI camera, said in the Orbis news release. "It cannot take a long time at all to learn how it works."

Orbis's new study is a follow-up to one the organization published in 2022 showing that AI screening led to patients' increased uptake of diabetic retinopathy referral services.²

Reference:

1. Whitestone N, Nkurikiye J, Patnaik JL, et al. Feasibility and acceptance of artificial intelligence-based diabetic retinopathy screening in Rwanda. British Journal of Ophthalmology. Published Online First: 04 August 2023. doi: 10.1136/bjo-2022-322683

2. Wanjiku Mathenge, MD, PhD, Noelle Whitestone, John Mkurikiye, MD, David H. Cherwek, MD, Nathan Congdon, MD, MPH, Nicolas Jaccard, PhD. Impact of artificial intelligence assessment of diabetic retinopathy on referral service uptate in a low-resourcing setting. Ophthalmology Science. Published April 22, 2022. DOI: https://doi.org/10.1016/j.xops.2022.100168