UWI Engineering Student Develops Model to Detect Diabetic Retinopathy
By: , June 17, 2025
Photo: Rudranath Fraser
Minister without Portfolio in the Office of the Prime Minister with responsibility for Efficiency, Innovation, and Digital Transformation, Senator the Hon. Ambassador Audrey Marks, observes as Engineering Student, Ricardo Harrison, explains his project, titled ‘Diabetic Retinopathy Using Computer Vision’, during the inaugural Mona Engineering Project Expo, held at the University of the West Indies, Mona, in St. Andrew on June 5.The Full Story
Final-year engineering student at the University of the West Indies, Mona, Ricardo Harrison, has developed a model to detect diabetic retinopathy (DR), using artificial intelligence (AI)-powered computer vision.
In an interview with JIS News, he said the solution is intended for low-resource hospitals, as the typical diagnostic machines are large, expensive and unportable.
He further shared that DR is a leading cause of blindness, especially in the Caribbean, where access to specialised healthcare is limited.
“A lot of Jamaicans and Caribbean people, in general, tend to not be aware that the amount of sugar that they intake tends to affect their vision later on. So, what my project really aims at doing is to provide a solution for hospitals to detect diabetic retinopathy at a more cost-effective way,” Mr. Harrison said.
His project improves upon existing solutions by offering a lightweight, locally deployable system with a user-friendly interface designed for non-specialist healthcare workers.
“When the enclosure is placed over the head, there is a 20D lens on the inside. So, what the 20D lens does is that it zooms in on the retina. It has a magnification of 3.3 times,” he explained.
Thereafter, using the mobile application he created, a photo of the retina is taken using a phone camera.
“There is an AI model that I created that I hosted on Google. So, after the photo is taken, it shows in the app. If the photo is clear enough, the app will say it’s clear enough and you can go ahead and upload. If it’s not clear enough, it would prompt you to retake another photo. So, if it’s clear enough, you click on upload and upload it to the server,” he detailed.
The server then analyses the image and picks out certain details in the image to tell whether it shows signs of diabetic retinopathy.
It tells the level of severity – mild, moderate, proliferative or severe – and returns a diagnosis to the app.
“It takes time off the healthcare professional’s hand and it reduces the money being spent on very expensive devices,” Mr. Harrison told JIS News.
To further support diabetic healthcare, the system also includes a patient monitoring feature that tracks heart rate and body temperature, providing a more comprehensive health assessment.
Based on the performance of the entire system, Mr. Harrison said it is approximately 78 per cent ready for integration in a clinical setting.
The project was guided by proposers from the University of Leeds in the United Kingdom, Gerardo Loza and Nikita Greenidge, and supervised by Head of Biomedical Engineering at UWI Mona, Sasha-Gay Wright.