Artificial Intelligence Model Predicts Age Using Retinal Images
An artificial intelligence developed by researchers at Seoul National University calculates people's age by analyzing retinal images with a low margin of error.
Scientists from Seoul National University and Seoul National University Hospital in South Korea have developed a new artificial intelligence model capable of predicting individuals' age from retinal images with an average margin of error of 2.5 years. Published in the journal GeroScience, the study revealed the relationship between retinal age differences and various health conditions.
Model Developed with Retinal Images
The retina, located at the back of the eye, holds the potential to offer new clues regarding aging. Scientists from Seoul National University and Seoul National University Hospital designed an artificial intelligence model that analyzes retinal images.
Amount of Data Used in the Research
The comprehensive study, published in the peer-reviewed scientific journal GeroScience, utilized 29,530 retinal images obtained from a total of 7,535 individuals. Scientists conducted detailed analyses using these data.
Margin of Error Rate of Artificial Intelligence
In tests conducted on different groups, the developed artificial intelligence model was determined to be able to predict people's real age with an average margin of error of 2.53 to 2.66 years.
Effect of Diabetes and Smoking
The retinas of individuals diagnosed with diabetes appeared an average of 2.52 years older than those of healthy individuals. Significant increases in retinal age differences were also detected in active smokers and former smokers.
Cataracts and Macular Degeneration
The estimated retinal age of eyes with cataracts was calculated to be an average of 1.86 years higher, while for individuals with age-related macular degeneration, it was calculated to be approximately 0.60 years higher.
Role of Vascular and Nervous Tissues
The retina is among the few areas in the body where both vascular and nervous tissues can be directly imaged, a condition that holds importance in monitoring aging and diseases.
Warnings from Scientists
Researchers emphasized that the results should not be evaluated directly as a biological age test, stating that the method can be used for now in examining public health disparities.