737 Deep Learning Methods Identify Eyelid Laxity As the Main Feature Causing the Aging Look

F. Wang,Z. Li,X. Hu,R. Ye,L. Du,S. Wang
DOI: https://doi.org/10.1016/j.jid.2022.05.749
2022-01-01
Abstract:The aging look has become a widespread concern which could be potentially related to one’s health status and well-being. Previous studies have tried to identify extrinsic and intrinsic risk factors of the aging look. In this study, we focus on the main facial aging signs which caused the aging look. Firstly, by carrying out the eye movement experiment and deep learning visualization in 26 volunteers, we identified orbit is the main region contributing to the aging look. Then, using deep learning methods, we quantified four periorbital aging features (i.e., laxity eyelids, pouch, crow’s feet and lacrimal sulcus) in the Eve cohort (n=255,578). We showed that eyelid laxity is the major factor for the aging look, compared with the other periorbital aging signs (β = 0.42, P = 0.01). Moreover, a stratification analysis by age showed that the weight of eyelid laxity is highest in all age groups (0.31-0.42), while the weights of lacrimal sulcus and eyes bag increased after 40 years old (0.14-0.24 and 0.07-0.18, respectively). At last, in a Han Chinese cohort (n=5,032) with data of eyelid laxity as well as the genome and environmental factors, we found eyelid laxity is significantly associated with environmental factors (i.e., sun exposure and air conditioner usage), but not with any genetic locus. This indicates that the periorbital aging features are probably affected by genes with minor effects, while environmental factors play a more prominent role.
What problem does this paper attempt to address?