20.10.2026 17:01
In research conducted in the US, with preliminary findings published in Lancet Digital Health, it was claimed that an artificial intelligence system called 'FaceSurvival', trained with millions of healthy faces and photographs of 35,000 cancer patients, could predict patients' survival time by looking at the areas under the eyes and around the bridge of the nose. The researchers stated that there is still some way to go before the system can be used in hospitals.
A study conducted in the US argued that artificial intelligence can be used to assess how advanced a disease is from a patient's facial photograph.
Scientists announced that the AI-supported system was trained on millions of healthy faces and photographs of 35,000 cancer patients. By closely examining features around the eyes, this system can predict survival times.
Researchers also noted that there is still some way to go before this technology is suitable for use in hospitals. They stated that their aim is to help recommend the right treatment to patients. The preliminary findings of the study were published in the well-known medical journal Lancet Digital Health.
THE EYES SAY IT ALL
The system in question is called 'FaceSurvival'. When estimating how much life a person has left, the technology focuses on the areas under the eyes and around the bridge of the nose and suggests that these are the most critical areas.
These criteria were applied to many cancer types at different levels, including cancers that had spread to the patient's bones or brain. The cancer patients in the study, most of whom were American and Dutch, were in their 50s or 60s. There were also children and patients in their 90s among them.
How ill people undergoing treatment were and what course their disease might take were already being monitored through examinations, comprehensive check-ups and tests. FaceSurvival, whose patent application is pending, is being developed as a technology to assist doctors.
Dr. Raymond Mak from Mass General Brigham Hospital in Boston, who is the director of the study, describes their goal as follows: "As patients get older, physicians need better methods to determine physical frailty before recommending intensive treatments such as radiation, surgery or systemic therapy. Our aim is to make health assessment based on facial photographs more quantitative, consistent and scalable. Rather than replacing the physician's judgment, it is to enable it to support personalized treatment decisions."
IT ALSO CALCULATES THE AGE YOU LOOK
Mak and his colleagues are also working on an artificial intelligence system that can estimate a person's biological age, that is, the age they appear, compared with their chronological age.
This system, called 'FaceAge', works on the assumption that the skin below the nose and near the mouth (also known as nasolabial folds and puppet lines), the forehead (frown lines), the temples (crow's feet) and the cheeks are a good indicator of how quickly we are aging. Some habits such as excessive alcohol consumption, drug use and smoking can accelerate this process.
Mak says the system produces "surprisingly consistent" results in a person's makeup-free photographs: "A stable result was also obtained in most of the cosmetic surgical procedures tested; only a small change was observed after rhinoplasty (nose surgery)."
EXPERTS ARE CAUTIOUS
Researchers emphasize that they need much more data and facial photographs, especially from different ethnic groups, to determine how well their tools work. These systems are still in the development stage. The research team also invited the public to upload their facial photographs to receive personalized biological age estimates.
Professor Miroslaw Bober, an expert in video processing and artificial intelligence systems at the University of Surrey in the UK, pointed out that it is still too early to talk about the capacity the system will reach in the future. Bober said, "This is a promising idea, but it is not yet ready for clinical use," and said the system needs to be tested by doctors according to patient groups.
Bober issued the following warning: "Predicting an outcome is not the same as identifying it."
We used artificial intelligence in the translation of this article, whose original is in English. A BBC journalist checked the translation before publication.