What if a simple photo of your face could predict your chances of surviving cancer? This is the Ambition of Facing, a new AI tool developed by researchers from the Mass General Brigham.
Facing does not measure not the chronological age, but the biological age visible through certain areas of the face. He analyzes subtle features, such as the muscle mass of the temples or the relaxation around the eyes. Then he tries to assess a patient’s ability to withstand heavy treatment.
THE Dr Ray Makradio-oncologist at Brigham and Women’s Hospitalco-author of the study, explains that doctors often have no choice but to trust their intuition. When a patient seems older than it is, some treatments are not offered to him. Conversely, a person in good shape despite their age could benefit from further care, but doctors still hesitate too often.
The example of a patient treated beyond standards
Jay Ball86, marked the medical team with its unusual vitality. Affected by lung cancer, he received more aggressive radiotherapy than what is normally proposed at his age. Four years later, Ball is still going well and faceage confirmed his biological age: About ten years younger than its real age.
For Ball, this result does not really surprise. It attributes its form to A Long Happy Weddinga taste for puns, technological curiosity and A lonked family. His parents exceeded 90 years and an aunt went up to 103.
To develop faceage, researchers used a public database with More than 56,000 photos of healthy people. They compared these faces to those of more than 6,000 patients with cancer, photographed just before their treatment. Result: sick patients appear on average Five years older than their real age.
Hugo Aertsco-author of the study, specifies that faceage allows Detect patients at high risk. People estimated at over 85 biologically are those who have the shortest survival. Conversely, more than 75 % of patients estimated at under 65 are still alive five years later.
What the tool really looks at a photo
Contrary to what one might think, faceage does not stop with white hair or front wrinkles. He especially analyzes Loss of muscle mass Around the temples, the hollows in the eyes and the finesse of the fabrics around the nose. Mak specifies that These areas change with agemuch more than other aesthetic benchmarks. It is these micro-signs that influence the facial estimate by AI.
Alpa Patel, of The American Cancer Societyrecognizes the faceage potential but invites caution. Many factors can influence the image analyzed: Light, makeup, surgery or even camera quality. According to her, the tool must still be validated and supervised before being used in the clinic. Mak shares this opinion. His team will soon launch a clinical trial with 64 patients with lung cancer. The objective is to compare faceage estimates to those of doctors, while analyzing complementary needs such as physiotherapy.
AI is slowly essential in modern medicine
Faceage arrives in a period when AI is already very present in hospitals. She analyzes mammograms, reads ophthalmic images and class data in electronic files. Similar tools today make it possible to summarize months of medical history in just a few hours.
But these progress is also accompanied by challenges. Andrew Beamresearcher and editor at the NEJM AI, alert on Cultural, sexist or racial biases that AI can reproduce and strengthen. The faceage team claims to have used various data, but continues to adjust its model to limit biases. She also works to improve precision on faces of cancer patients who have undergone aesthetic changes.
A more ambitious question is already agitating the Dr. Mak team. One day, we can reverse the clock of aging thanks to the image ? Can we extend life by observing the evolution of facial features? It is not yet the faceage clinical priority, but the idea already nourishes the next research.
.