Individual
senescence refers to time change, mainly decline, of the ability of the body to cope with physiological needs.
Biological age (BA) is a biomarker of sequential aging and can be used for demographic stratification to predict certain age-related chronic diseases. The BA may be predicted from biomedical features such as
brain MRI,
retina or facial images, but heterogeneity inherent in the aging process limits the usefulness of predicting the BA from the individual
body system. The methods disclosed herein teach a multi-
modal Transform-based architecture with cross attention that can combine face, tongue, and
retinal images to estimate the BA. The model is trained using face, tongue, and
retina images from 11,223
healthy subjects and demonstrates that fusion using the three image modalities achieves the most accurate BA prediction. The methods are validated on a test
population including 2, 840 individuals suffering from six chronic diseases and result in a more
significant difference between timing age (CA) and BA (AgeDiff) compared to
healthy subjects. AgeDiff can be used as a separate biomarker or in combination with other known factors for
risk stratification and progression prediction of
chronic disease. Thus, these results emphasize the feasibility of using multi-
modal images to estimate and explore the aging process.