Deep learning aging marker from retinal images unveils sex-specific clinical and genetic signatures
Researchers fine‑tuned the RETFound foundation model on 71,343 UK Biobank retinal fundus images, achieving a mean absolute error of 2.85 years in age prediction. The derived retinal age gap correlates with cardiometabolic traits, inflammation, cognitive performance, all‑cause mortality, dementia, cancer, and incident cardiovascular disease, and genome‑wide association studies identified genes linked to longevity, metabolism, neurodegeneration, and age‑related eye diseases. Sex‑stratified analyses show consistent model performance but distinct biological signatures: males exhibit stronger link…
Researchers fine‑tuned the RETFound foundation model on 71,343 UK Biobank retinal fundus images, achieving a mean absolute error of 2.85 years in age prediction. The derived retinal age gap correlates with cardiometabolic traits, inflammation, cognitive performance, all‑cause mortality, dementia, cancer, and incident cardiovascular disease, and genome‑wide association studies identified genes linked to longevity, metabolism, neurodegeneration, and age‑related eye diseases. Sex‑stratified analyses show consistent model performance but distinct biological signatures: males exhibit stronger link…
The study showcases the use of foundation models for non‑invasive aging biomarkers, supporting Swiss AI efforts in precision medicine and health data analytics.
Original source record. Open the original record to verify the underlying announcement.