Researchers from the Buck Institute for Aging Research, Google Health and Zuckerberg San Francisco Hospital
Minor changes in retinal capillariesoften go unnoticed, but they can be detected using a trained neural network, the scientists explain. Previously, Google researchers developed models to predict diabetic retinopathy from retinal images and used them to detect at least 39 eye diseases.
To estimate age, the researchers trained thismodel based on a data set that includes dynamic observations for more than 100 thousand patients. The finished model was tested on retinal images of 64 thousand people obtained from the British biobank.
The study showed a high correlation betweeneyeAge estimated age and chronological age. However, the results show a more accurate positive predictive rate for two consecutive visits to individuals, rather than for random, age-matched individuals.
The results show that potentially less thanin a year, we will be able to determine the aging trajectory with 71% accuracy, noting noticeable changes in the eyes of those undergoing treatment, providing a valid assessment of geroprotective therapy.
Sara Ahadi, co-author and former Google Research Fellow
Researchers note that currentlyAging assessment uses a phenotypic age assessment method that is based on the analysis of blood biomarkers. This method has been widely used, but it is not well suited for assessing short-term changes. On the contrary, the thinnest blood vessels in the retina record all changes in the body and do not depend on food intake or infectious disease. Together with other tools, they can be used to assess the effects of anti-aging therapies.
Read more:
Biologists discover how cancer cells elude the immune system
The key theory of quantum physics has finally been proven. Main
Named products that protect the brain from dementia, and when to use them