FaceSurvival AI Estimates Cancer Survival From a Photo — SkimNews
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- FaceSurvival was trained on millions of healthy faces and 35,000 cancer-patient photos to estimate survival across multiple cancer types and levels of severity, including cancer that had spread to bones or the brain.
- FaceSurvival suggested that the area under the eyes and around the bridge of the nose was most useful for survival estimates; most patients came from the US and Netherlands, were in their 50s or 60s, and ranged from children to people in their 90s.
- FaceSurvival remains a long way from hospital use: its preliminary findings appeared in Lancet Digital Health, a patent application is pending, and the researchers say more facial photos are needed, particularly from different ethnic groups.
- Raymond Mak said FaceSurvival is intended to quantify reserve and frailty before radiation, surgery or systemic therapy, supporting personalized treatment decisions rather than replacing physician judgment.
- FaceAge was designed to estimate biological age using skin below the nose and near the mouth, forehead frown lines, temple crow's feet and cheek features; the researchers reported consistency across real-world photos despite pose, lighting, expression and makeup, with only a small shift after rhinoplasty.
- Miroslaw Bober called the work a promising proof of concept that is not ready for clinical use, said claims about tracking ageing or lifestyle changes over time were premature, and stressed that prediction is not the same as explanation.
Why it matters: For doctors and cancer patients facing treatment choices, FaceSurvival is designed to add a consistent, scalable assessment to existing examinations rather than displace physician judgment. Its 35,000 cancer images are promising, but greater ethnic diversity and testing across hospitals and patient groups are still needed before clinical use.
Ask SkimNews
