New AI Tool Estimates Biological Age Using Selfies, Aiding in Personalized Healthcare

Researchers at Mass General Brigham in Boston have developed a new artificial intelligence tool, FaceAge, that uses facial photographs—such as selfies—to estimate a person’s biological age, a measure of health that may differ significantly from chronological age, the Washington Post reports.

Unlike a person’s actual age in years, biological age reflects an individual’s physiological state and overall health. The development of FaceAge, detailed in The Lancet Digital Health, is part of a broader effort to enhance medical assessments and treatment decisions through AI-driven tools.

FaceAge was trained on a dataset of approximately 59,000 images of healthy individuals aged 60 and older, many of which were sourced from public platforms like Wikipedia and IMDb. The algorithm analyzes facial features to estimate biological age and has already been tested on over 6,000 cancer patients. Results showed that many patients had a biological age significantly higher than their chronological age, and this disparity often correlated with poorer survival outcomes.

In early testing, FaceAge improved doctors’ ability to predict life expectancy. When given only a photo, physicians correctly estimated six-month survival 61% of the time. That accuracy increased to 74% with clinical information added, and rose to 80% when using both clinical data and FaceAge.

Biological age is considered important in determining suitable treatment options. For example, in cancer care, older patients may not always be considered for aggressive therapy due to concerns about frailty. FaceAge could serve as an objective tool to support or challenge such decisions.

Dr. Raymond Mak, a radiation oncologist and one of the lead developers, shared the story of an 86-year-old patient with terminal lung cancer who responded well to aggressive treatment. Later analysis with FaceAge showed the patient’s biological age was more than a decade younger, potentially justifying the decision to treat more intensively.

Researchers emphasize that FaceAge is not intended to replace clinical judgment, but rather to complement it. Experts unaffiliated with the study, including AI specialists from the Mayo Clinic and City of Hope, called it a promising early-stage tool that could eventually serve as a new biomarker—pending further validation and regulatory approval.

The algorithm must still overcome challenges such as biases related to lighting, makeup, and cosmetic surgery. Developers are working to improve the model’s robustness and account for variations across different populations and image conditions.

Ethical concerns, particularly regarding privacy and misuse by non-medical parties such as insurers, remain a priority for the researchers. They stress the need for strict governance if the tool is eventually adopted in clinical settings.