A study found that facial photos kept in hospitals for treatment records could serve as clues to reading health changes in cancer patients. The researchers said more validation is needed, but noted the photos could potentially be used as an auxiliary indicator to monitor treatment progress without burdening patients.
Researchers at Mass General Brigham said that using an AI tool called "FaceAge," which estimates biological age from a person's facial photo, they found that the rate of facial aging in cancer patients is associated with survival. The findings were published in the international journal Nature Communications on the 28th.
Biological age is a concept that reflects actual physical condition and the degree of aging. Even among people who are 60, some are physically younger while others age faster. FaceAge estimates biological age by using Deep Learning to analyze various features in facial photos such as wrinkles, skin condition, and facial contours.
Previously, the team reported that cancer patients tend to look about five years older than their chronological age on average when analyzed with FaceAge, and that patients with higher biological age often had poorer survival after cancer treatment.
In this study, the researchers analyzed facial photos of 2,279 cancer patients who received radiation therapy at least twice at Brigham and Women's Hospital from 2012 to 2023. The photos were not taken for experimental purposes, but as part of routine procedures during radiation therapy.
Comparing two photos taken at different points during treatment to calculate the rate of facial aging, the researchers found that patients' faces aged on average 40% faster than actual time. In particular, patients with faster facial aging had lower survival prospects. The association was more pronounced when the interval between photos was two years or longer.
The researchers also analyzed an existing metric, the "age gap." The age gap refers to how much the biological age seen in a single photo exceeds chronological age.
The analysis showed that patients with high values for both the age gap and the facial aging rate tended to have worse prognoses. However, for predicting long-term outcomes, the facial aging rate, which tracks change over time, provided more stable information than the age gap derived from a single photo.
The researchers said the technology could become a "noninvasive biomarker" that does not burden the body like blood tests or biopsies. A biomarker is a biological signal that indicates disease status or treatment response. In particular, facial photos are easy to obtain during hospital visits and carry a low expense.
However, the team added, "For FaceAge and the facial aging rate to be widely used in actual clinical settings, additional validation is needed across more diverse races, ages, and disease groups," and, "Follow-up studies are examining whether they can be applied to various cancers, chronic diseases, and health status assessments."
References
Nature Communications (2026), DOI: https://doi.org/10.1038/s41467-025-66758-w