AI tool claims your selfie can predict cancer survival more accurately

FaceAge, an AI tool developed by Harvard researchers, estimates biological age using facial images and can predict cancer survival outcomes more accurately than doctors

What if your face could reveal more about your health than your actual age? Scientists have developed a new artificial intelligence (AI) tool called FaceAge, which uses just a photo of your face to estimate your biological age—an indicator of how your body is ageing. Published recently in The Lancet Digital Health, the study titled “FaceAge, a deep learning system to estimate biological age from face photographs to improve prognostication” found that FaceAge not only measures ageing more precisely than a birthday count but also predicts survival chances in cancer patients better than doctors do.

What is FaceAge and how does it work?

FaceAge is a deep learning system developed by researchers at Harvard Medical School and partners. It analyses simple face photographs to predict a person’s biological age, which can be different from their chronological age. Unlike a traditional age number, biological age reflects the wear-and-tear on your body due to genetics, lifestyle, disease, and environment.

 

 

The AI model uses face detection and feature extraction techniques to estimate how old someone looks biologically. It was trained on over 58,000 face images of healthy people and tested on 6,196 cancer patients across multiple hospitals in the US and Europe.

FaceAge predicts cancer survival more accurately than doctors

According to the paper, in three major cancer patient groups—curative (patients receiving curative treatments, primarily radiation therapy), thoracic (patients with cancers in the chest area), and palliative (patients with advanced-stage cancer)—FaceAge was a stronger predictor of survival than chronological age, even when doctors already had full clinical information.

 

In patients receiving curative cancer treatments, higher FaceAge scores correlated with significantly worse survival outcomes. For those with metastatic cancer receiving palliative care, FaceAge improved survival predictions when added to existing clinical tools, like the TEACCH  (Treatment and Education of Autistic and Communication-related handicapped CHildren) model.

 

When doctors were shown FaceAge data alongside patient records, their ability to predict six-month survival improved substantially. 

Why biological age is more useful than chronological age

Chronological age doesn’t always match how healthy or frail someone actually is. For instance, a 65-year-old who exercises regularly and eats well might have the body of a much younger person, while a 28-year-old who smokes and eats poorly might age faster internally. FaceAge fills this gap by identifying biological ageing signs invisible to the naked eye.

 

In this study, cancer patients looked nearly five years older than their real age, on average. This facial age difference was linked to poorer survival outcomes, making FaceAge a potential game-changer in clinical settings.

 

FaceAge was also found to be genetically associated with senescence genes—linked to the process of cellular senescence—which plays a role in ageing and cancer development. Chronological age showed no such correlation.

FaceAge could transform cancer treatment—but not yet

Researchers believe FaceAge could help doctors make more objective treatment decisions, particularly in cancer care, where risks and benefits must be weighed carefully. It could also support personalised medicine and clinical trials by offering a consistent measure of patient health.

 

However, the tool still requires further validation in more diverse populations before it can be adopted widely in clinical practice. 

Ethical concerns around AI and facial health prediction

The use of AI tools like FaceAge raises ethical and privacy concerns. Researchers warn of potential misuse by insurance firms or employers to make decisions about individuals based on facial data. There’s also uncertainty about whether FaceAge performs equally across races, age groups, and genders—or if results could be skewed by facial features, cosmetic surgeries, or image quality.

 

The authors call for strong regulation and additional research to ensure AI tools in healthcare are fair, transparent, and safe.  For more health updates and wellness insights, follow #HealthWithBS