AI analysis of eye scans can detect diseases such as diabetes, osteoporosis and thyroid disease in seconds

A new study presents an artificial intelligence system that scans images of the retina to detect signs of diabetes, high blood pressure, high cholesterol, gout, osteoporosis and thyroid disease in seconds. The program—called Reti-Pioneer—is a step toward being able to diagnose many different conditions from a scan of the eye, providing people a quicker diagnosis for common conditions and increasing access to crucial testing.

Associate Professor Lisa Zhuoting Zhu, head of ophthalmic epidemiology at CERA, is one of the leading authors on the paper published in Nature Medicine. She says this technology is making disease diagnosis more efficient, particularly in remote or regional communities.

"This technology will be a real benefit to public health," says Zhu. "Patients would be able to get information about their health instantly and start interventions as soon as possible instead of waiting for more time-consuming test results."

How the eye scan works

Reti-Pioneer uses several AI systems to scan images of the eye and check for subtle signs of disease that would be impossible to spot otherwise. These models are trained using thousands of images of eyes from people both with and without disease. They are taught to spot signs of illness that can't be picked up by the naked eye.

A scan of the eye is not only cheaper and easier to perform than forms of diagnosis like a blood test, but it also delivers results in seconds.

Faster screening in clinics

The research, primarily carried out in primary health care centers in China, found the system provides accurate screening to help doctors make decisions about patients' health without waiting for these slower test results.

"China has one of the most efficient health care systems in the world—the results of a blood test can come back to patients in nine hours," Zhu says. "But this system is even quicker.

"In countries like Australia, Singapore and the United States, test results can take a few days—or even a week if samples need to be taken to a lab from remote communities.

"If a patient can be flagged on the spot for a condition like diabetes, we can start interventions while waiting for the results of more advanced screenings that take time."

Potential in remote communities

Another advantage of the system is that it requires only a basic fundus camera, rather than specialized equipment. This means it could be used in GP clinics, optometrists' offices, community pharmacies and traveling clinics.

Zhu says systems like these have real potential to close access gaps. "AI platforms like this can significantly increase access to health care, particularly for those in regional and remote Aboriginal communities."

This article was originally published on MedicalXpress Breaking News-and-Events.