by Jonathan van Bilsen
Imagine a tool which can calculate the age your face reveals; a new method which could someday revolutionize preventive health and aging research. Jing-Dong (Jackie) Han, a computational biologist, at Peking University, has developed an AI-driven “facial aging clock” which interprets age, based on a detailed 3D image of the face. By analyzing these images, the system calculates a person’s “biological age,” which might differ significantly from their chronological age. Unlike simply counting years, biological age gives a sense of the wear and tear on cells and tissues, hinting at a person’s overall health.
Ms. Han’s work draws inspiration from a longstanding Chinese tradition of face-reading, to assess health. To build her model, Dr. Han’s team gathered 3D facial images, from about 5,000 individuals, in Jidong, China. They crafted two AI models: one for chronological age and another for biological age, each tracking facial changes which naturally happen over time. The drooping of the eyes, the widening of the nose, and a sagging jawline aren’t just cosmetic shifts, they can also reflect deeper health issues. For instance, systemic inflammation has been linked to sagging skin.
According to Andre Esteva, CEO of a medical AI start-up, this new tool could shift how doctors approach patient care. A quick facial scan, to assess biological age, could help individuals make lifestyle changes or assist doctors in monitoring patients undergoing challenging treatments. This technology also has research potential, allowing scientists to track how interventions might affect aging.
AI tools depend on a wealth of data, where the ‘correct’ answer, or ‘ground truth’ is, known in advance. For biological age, however, no definitive standard exists. Christopher Bell, a researcher at the University of London, explained, biological age is based on many age-related changes in our cells and tissues. For example, telomeres (like the plastic tips on shoelaces) protect chromosome ends from fraying, while mitochondria (tiny power plants inside cells) produce the energy cells need to work properly.
The journey to understanding aging began with DNA clocks, where researchers noted how specific DNA patterns, affected by chemical changes, could hint at cellular aging rates. Other clocks have examined protein levels in the blood or stem cell division frequency. In 2016, Jackie Han pivoted to using appearance as an age indicator, inspired by a 2009 study, where the older-looking twin often had poorer health outcomes.
As her research expanded, Dr. Han harnessed AI to analyze facial features on a larger scale. Her team assessed biological age by collecting independent estimates, from five human observers, for each subject. The AI learned from these evaluations and became impressively accurate, generally predicting ages within three years of actual values.
This model also identified ‘fast-agers’ and ‘slow-agers’—people whose facial age differed by more than three years from their chronological age. For Dr. Jing-Dong Han, these outliers reveal the most about the aging process and may someday point the way toward healthier aging paths.
To summarize, a tool will soon be able to scan our face, combine this with our age and reflect potential health risks which may be avoidable. For me, I just want to look twenty again.
Jonathan van Bilsen is a television host, award winning photographer, published author, columnist and keynote speaker. His show, ‘The Jonathan van Bilsen Show’, on RogersTV, the Standard Website or YouTube, features many of the people included in this column.
Comments