A Deep Learning-based Indicator to Reveal Biological Age Using Lens Photographs
Overview
- Phase
- Not Applicable
- Intervention
- Not specified
- Conditions
- Ophthalmology
- Sponsor
- Sun Yat-sen University
- Enrollment
- 6000
- Locations
- 1
- Primary Endpoint
- The difference between LensAge and chronological age
- Status
- Recruiting
- Last Updated
- 3 years ago
Overview
Brief Summary
Assessment of aging is central to health management. Compared to chronological age, biological age can better reflect the aging process and health status; however, an effective indicator of biological age in clinical practice is lacking. Human lens accumulates biological changes during aging and is amenable to a rapid and objective assessment. Therefore, the investigators will develop LensAge as an innovative indicator to reveal biological age based on deep learning using lens photographs.
Investigators
Haotian Lin
Professor
Sun Yat-sen University
Eligibility Criteria
Inclusion Criteria
- •ages from 20 to 100 years
- •have anterior segment photographs
- •have ophthalmic and physical examination records
Exclusion Criteria
- •have a history of previous eye surgery, eye trauma, or ocular diseases that can cause complicated cataracts
- •baseline information missing
Outcomes
Primary Outcomes
The difference between LensAge and chronological age
Time Frame: Baseline
The age estimation models based on a convolutional neural network (CNN) using lens photographs will be used to generate LensAge. LensAge at the individual level will be calculated by averaging the results of all images corresponding to one individual. The difference between LensAge at the individual level and chronological age will be used to unveil an individual's aging process. A difference above 0 indicates an individual with a faster pace of aging than their peers of the same chronological age, while a difference below 0 indicates a slower pace of aging.
Secondary Outcomes
- Correlation between the LensAge difference and age-related health parameters(Baseline)
- Mean error (ME) of the DL age estimation model.(Baseline)
- Mean absolute error (MAE) of the DL age estimation model.(Baseline)
- R-squared (R2) of the DL age estimation model.(Baseline)