
The study involved over 59,000 facial images and data from 6,200 oncology patients. It was found that cancer patients appeared, on average, five years older than their actual chronological age. This allowed the AI to make accurate predictions about biological aging and survival rates.
"How old a person looks compared to their actual age truly matters in evaluating health" said Hugo Aerts, Director of the AI in Medicine program at Mass General Brigham.
Estimating biological age through facial photos
Predicting survival rates in severe illnesses
Supporting palliative care and medical decision-making
Beyond oncology, the tool may be applicable to diagnosing age-related chronic diseases.
However, there are limitations. The model was trained primarily on light-skinned individuals, and factors such as lighting, makeup, and cosmetic treatments may affect accuracy. Researchers plan to expand the dataset and test FaceAge on more diverse patient groups.
Keywords