NEW DELHI, India — A team of researchers from India and the United States has developed an artificial intelligence-based technique that can detect diabetes without the need for traditional blood tests, using high-resolution images of the retina.
The method analyzes photographs of the back of the eye to identify signs of elevated blood sugar levels. According to the researchers, the AI system can detect subtle changes in retinal blood vessels that are not visible to the human eye, allowing it to distinguish between people with and without diabetes.
The findings were published in the journal Diabetes Technology and Therapeutics.
“India has over 100 million people with diabetes, and very often many do not even know they have it,” said Dr. V. Mohan, a Chennai-based diabetologist and Padma Shri awardee who was part of the study. “If the use of AI tools with simple retinal photos can help early diagnosis of diabetes, it can be used in real time in the future to screen for diabetes.”
Dr. Sudeshna Sil Kar of Emory University said the researchers trained the AI system to identify specific shapes and patterns in retinal blood vessels by analyzing images from people both with and without diabetes.
The research team, which also included scientists from Yenepoya University in Karnataka, analyzed 273 retinal images from 139 participants. Using machine vision-based techniques, they extracted 226 quantitative vessel tortuosity features from arteries and veins.
In testing, the AI system identified diabetes with 95 percent sensitivity using retinal photographs. It was also able to detect prediabetes, a stage in which lifestyle changes can help prevent the onset of the disease.
Researchers said the technique could offer an effective non-invasive option for early diabetes detection, particularly in settings where access to laboratory testing is limited. The approach does not require fasting, blood samples, or expensive equipment, relying instead on a quick retinal photograph.
The team emphasized, however, that the findings will need to be validated through studies involving larger and more diverse populations before the technology can be widely deployed. (Source: IANS)
