FaceAge: The AI that analyses your selfie to predict cancer risk — would you try it?

Dubai: A groundbreaking study published in The Lancet Digital Health has unveiled an AI tool that can analyse facial features from a selfie to predict how fast a person’s body is aging — offering doctors a new way to estimate cancer risk and tailor treatments.

Developed by scientists at Mass General Brigham, the deep-learning algorithm, called FaceAge, was trained on nearly 60,000 photos of healthy individuals and later tested on over 6,000 cancer patients beginning radiotherapy.

Unlike standard age estimates, FaceAge doesn’t rely on birth certificates. Instead, it evaluates subtle signs — such as skin texture, eye shape, and facial muscle tone — to estimate biological age, a number that reflects how worn or resilient a body truly is.

On average, cancer patients in the study appeared about five years older than their chronological age. Each additional year of biological age correlated with a significant reduction in survival, making it a powerful new predictor.

For decades, doctors have used facial cues informally to judge a patient’s fitness for treatment. A frail appearance might prompt gentler therapy, while a vigorous look could justify more aggressive options. But such impressions are subjective.

FaceAge offers an objective, data-backed estimate. “We can use AI to estimate a person’s biological age from face pictures,” said lead researcher Hugo Aerts. “Our study shows that information can be clinically meaningful.”

To test the tool’s real-world potential, researchers asked ten clinicians to predict whether patients receiving palliative radiotherapy would survive for several months. Even with access to patient records and chronological age, their accuracy was little better than chance.

When FaceAge was added to the mix, predictions improved significantly. This suggests the AI can pick up on subtle signs that human eyes often miss — potentially becoming a key ally in complex care decisions.

FaceAge could also prove useful in other medical fields. Since accelerated aging underlies many chronic illnesses — from heart disease to dementia — a tool that detects it early could enable preventive care years before symptoms arise.

“This opens the door to a new realm of biomarker discovery,” said Ray Mak, co-senior author of the study. “The potential goes far beyond cancer care or predicting age.”

Despite its promise, FaceAge isn’t ready for clinics just yet. The model was trained on data from just two hospitals, and variations in lighting, photo quality, or cultural factors like makeup and skincare could skew results.

The team plans to validate the tool across larger and more diverse populations. Crucially, they stress the need for strong ethical safeguards, including patient consent and transparency around data use.

FaceAge is a glimpse into a future where a simple selfie might help guide medical care — turning everyday images into powerful tools for detecting disease and personalizing treatment.

Source: The Lancet Digital Health