US researchers have developed an artificial intelligence tool that can estimate cancer survival rates by analyzing a single photograph of a patient’s face. The preliminary findings, published in the journal Lancet Digital Health, suggest the technology could eventually help doctors tailor treatment plans, though the team emphasizes it is far from ready for hospital use.
FaceSurvival and Biological Age Tools
The research team created two distinct AI models. The first, named FaceSurvival, analyzes facial features to predict how long a patient might live with cancer. The scientists identified that the areas under the eyes and around the bridge of the nose provide the most critical data for these estimates. This model was tested across various cancer types, including those that had spread to the bones or brain.
The second model, called FaceAge, estimates a person’s biological age rather than their chronological age. This tool focuses on the skin below the nose, the nasolabial folds, the forehead, temples, and cheeks. Lead researcher Dr. Raymond Mak from Mass General Brigham in Boston stated that the FaceAge tool was surprisingly consistent across common real-world photos. It accounted for differences in pose, lighting, facial expressions, and makeup, and remained stable across most tested plastic surgery procedures, with only minor shifts observed after rhinoplasty.
Supporting Physician Judgment
Dr. Mak explained that the primary goal is to assist physicians in assessing physical reserve and frailty before recommending intensive treatments such as radiation, surgery, or systemic therapy. The team aims to make health assessments more quantitative and scalable to support personalized treatment decisions. The tool is designed to complement, not replace, the comprehensive checks and tests doctors already perform.
The training data for the survival model included 35,000 photos of cancer patients from the US and the Netherlands. Most participants were in their 50s or 60s, though the dataset also included children and individuals in their 90s. The researchers have filed a patent application for the technology.
Need for More Data
Despite the promising results, the researchers stress that much more data is required. They are calling for more photos from diverse ethnic groups to ensure the tools work accurately across different populations. The team is currently inviting the public to upload their own facial photos to receive a personalized biological age estimate as part of their ongoing data collection efforts.
Prof Miroslaw Bober, an expert in video processing and machine learning at the University of Surrey in the UK, cautioned that claims about tracking aging or lifestyle changes over time are premature. He described the work as a promising proof of concept but noted it is not ready for clinical use. Bober added that the tool still needs testing against doctors’ own judgments across different hospitals and patient groups, warning that predicting an outcome is not the same as explaining it.
Source: BBC
