# Could Your Face Reveal Your Cancer Survival Odds? AI Study Links Older Appearance to 21% Higher Death Risk > Source: > Published: 2026-10-20 06:56:50+00:00 # Could Your Face Reveal Your Cancer Survival Odds? AI Study Links Older Appearance to 21% Higher Death Risk ## Researchers say facial AI could eventually support cancer treatment decisions, but must not replace clinical judgement An artificial intelligence (AI) tool developed by researchers at Mass General Brigham can estimate cancer patients' [biological age](https://www.ibtimes.co.uk/ai-drug-rentosertib-reverses-biological-age-1818792) and mortality risk from a single facial photograph, according to a study published in The Lancet Digital Health. Using its FaceAge biological-age estimation algorithm, the researchers found that cancer patients whose estimated biological age was at least five years above their chronological age had a 21 per cent higher risk of dying during an average four-year follow-up. A separate algorithm, FaceSurvival, predicts mortality risk directly from facial photographs without first estimating biological age. Trained on more than 40 million facial photographs, the framework aims to quantify the visual assessments oncologists use to judge a patient's frailty and physical resilience. The researchers stressed that the technology remains experimental. ## Facial Age Linked to Survival Risk Patients whose estimated biological age was at least five years below their chronological age had a 17 per cent lower risk of death. On average, cancer patients also appeared about five years [biologically older than their chronological age](https://www.ibtimes.co.uk/early-onset-cancer-biological-ageing-england-1824004), the researchers reported in an earlier study. The team developed two algorithms from the same foundation model to assess different aspects of health. FaceAge estimates biological age, reflecting the body's functional condition rather than the number of years a person has lived. It analyses facial features including creases running from the nose to the corners of the mouth, the temples and the forehead above the eye sockets. FaceSurvival, by contrast, predicts mortality risk directly from a photograph. It focuses on the area beneath the eyes and around the bridge of the nose. In validation tests, more than half of the cancer patients placed in FaceSurvival's highest-risk category died within 90 days. In the lowest-risk group, the figure was 2 per cent. Patients in the top risk quartile were more than three times as likely to die as those in the lowest. ## Smoking and Other Factors Affect Facial Age The researchers said the model's predictions remained stable across changes in lighting, camera angle, facial expression and makeup, as well as most [plastic surgery procedures](https://www.ibtimes.co.uk/sharon-osbourne-questions-further-cosmetic-surgery-facial-indentation-1826175). Rhinoplasty produced only a minor shift in predictions. The algorithms were validated using cancer patient cohorts in the United States and the Netherlands, including datasets containing more than 38,000 and nearly 4,900 patients, respectively. [Smoking, heavy drinking and drug use](https://www.ibtimes.co.uk/ai-researchers-smoking-cancer-cure-palantir-cto-shyam-sankar-1825291) were strongly associated with older facial biological age. In earlier FaceAge research, current smokers appeared roughly three years older than former smokers and people who had never smoked, with no significant difference between the latter two groups. The researchers said those findings suggested facial characteristics could recover after smoking cessation. ## Researchers Warn Against Relying on AI Alone The technology could eventually help oncologists assess whether patients are likely to tolerate aggressive treatments such as chemotherapy, radiotherapy or surgery, or whether gentler palliative care may be more appropriate. However, the researchers said the algorithms are intended to support [personalised treatment decisions](https://www.ibtimes.co.uk/ai-assisted-hospital-billing-healthcare-spending-rise-1822687), not replace clinical judgement. Relying solely on an automated score could lead to ethically problematic decisions. Patients who look older could be undertreated, while frail patients who appear younger might receive treatments they cannot tolerate. It also remains unclear whether using the predictions to guide treatment decisions would improve survival. The researchers said extensive clinical trials and further validation studies are needed before the technology can be introduced into routine hospital care. © Copyright IBTimes 2026. All rights reserved.