The PhotoAgeClock was trained on 8,414 anonymised high resolution eye corner photos in order to predict human age
Scientists have developed a deep learned photographic biomarker in an effort to help identify human age.
Estonian computer vision and AI startup company, Haut.AI, in partnership with Insilico Medicine, which uses genomics and data analysis to discover age-related diseases, conducted the research to advance the accuracy of predicting age.
The PhotoAgeClock predictor was trained on 8,414 anonymised high-resolution eye corner photos - which was found to be the most age-relevant area - with an accuracy level of 2.3 mean average error.
CEO of Haut.AI Anastasia Georgievskaya said: “Development of reliable biomarkers of ageing based on photographic images of the skin and face has the potential to accelerate ageing research and help identify interventions that improve skin health and beauty on an individual level.
“The study of multiple skin conditions using AI and computer vision may change the very approach to skin care.
“The main value of PhotoAgeClock and the non-invasive biomarkers trained on skin imaging data is in estimation of the differential changes induced by the various interventions.”
The researchers hope these ageing biomarkers can be used to provide early detection for diseases or prevent their onset.
This could also help the development of biomedical interventions and skin care treatments for the individual to offer a more personalised experience.