Abstract
Purpose :
To further refine the deep learning technology for retinal age (biological age based on retinal images) and comprehensively explore the clinical values of retinal age gap (retina-predicted age minus chronological age) in terms of the risk of multiple age-related diseases.
Methods :
Retinal age prediction was refined using Temporal Fundus Images Enhanced Progressive Label Distribution Learning. Retinal age gaps were determined for patients who had no history of corresponding diseases at baseline. Diseases events were determined by data linkage to hospital records on admissions and diagnoses. Cox proportional hazards regression models were performed to evaluate the association between retinal age gaps and incident diseases.
Results :
The mean absolute error (MAE) reduced from 3.5 to 2.8 years after refinement. During the follow-up time, a total of 849 cases of mortality, 640 cases of cardiovascular diseases (CVD), 75 cases of end-stage renal diseases (ESRD), 1252 cases of chronic obstructive pulmonary disease(COPD), 514 of dementia and 4460 of cancers were identified among the participants without the corresponding diseases at the baseline. Table 1 shows that after adjusting for multiple confounding factors each 1-year increase in retinal age gaps was associated with a 5% increase in the risk of mortality, 6% increase in CVD, 11% increase in ESRD, 3% increase in COPD, 7% in dementia and 1% in cancers (HR=1.05, 95% CI: 1.03–1.07, P<0.001; HR = 1.06, 95% CI = 1.04-1.09, P = 0.011; HR = 1.11, 95% CI = 1.02-1.19, P = 0.002; HR = 1.03, 95% CI = 1.01-1.05, P < 0.001; HR = 1.07, 95% CI = 1.04-1.11, P = 0.007; HR = 1.01, 95% CI = 1.00-1.02, P < 0.001, respectively).
Conclusions :
We found the retinal age gap was associated with an increased risk of multiple incident age-related diseases highlighting the value of retinal age as a reliable ageing biomarker. It may serve as an informative tool to facilitate the risk assessment and personalized screening pattern of ageing.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.