A deep-learning algorithm that predicts a person’s biological age from low-cost face photos could help determine their survival after cancer treatment, a fascinating study has revealed.
The FaceAge tool offers an objective way to clinically use insights gained into a person’s biological age and physiological health from their physical appearance, and could be used for other conditions.
The study, in The Lancet Digital Health, revealed that cancer patients undergoing palliative care looked, on average, around five years older than their chronological age, and that looking older was linked with worse outcomes.
“This work demonstrates that a photo, like a simple selfie, contains important information that could help to inform clinical decision-making and care plans for patients and clinicians,” said co-senior author Hugo Aerts, PhD, director of the Artificial Intelligence in Medicine (AIM) program at Mass General Brigham.
“How old someone looks compared to their chronological age really matters—individuals with FaceAges that are younger than their chronological ages do significantly better after cancer therapy.”
People age at different rates, with genetic and lifestyle factors such as diet, stress, smoking, and alcohol influencing the aging process and affecting DNA methylation status, telomere length, and both gene and protein expression patterns.
Physical appearance can therefore offer more reliable insights into biological age and physiological health than chronological age, but it has traditionally been incorporated into medical judgements in a subjective and non-standardized fashion.
The researchers therefore set out to develop a deep learning system that could objectively estimate biological age from easily obtainable and low-cost photographs of a face.
FaceAge was trained on 58,851 photos of people on public datasets, who were presumed healthy and aged at least 60 years.
It was then tested in 6,196 cancer patients from two centers in the Netherlands and the United States using photographs that were routinely taken at the start of radiotherapy treatment.
The team found that, on average, cancer patients looked a significant 4.79 years older than their chronological age compared with patients who did not have cancer and were treated for conditions that were benign or precancerous.
When 10 clinicians and researchers were asked to predict short-term life expectancy from 100 photos of patients receiving palliative radiotherapy, their predictions were little better than the toss of a coin.
Performance was better when face photographs were supplied along with clinical chart information, but improved even further after clinicians were also provided with FaceAge information.
FaceAge is correlated with molecular processes of cell-cycle regulation and cellular senescence, supporting the concept of FaceAge as a biomarker that relates to biological aging.
“This opens the door to a whole new realm of biomarker discovery from photographs, and its potential goes far beyond cancer care or predicting age,” said co-senior researcher Ray Mak, MD, also from the AIM program.
“As we increasingly think of different chronic diseases as diseases of aging, it becomes even more important to be able to accurately predict an individual’s aging trajectory.
“I hope we can ultimately use this technology as an early detection system in a variety of applications, within a strong regulatory and ethical framework, to help save lives.”
