An AI tool that predicts how long a patient might live—just by looking at their face!
American researchers have developed an AI tool called "FAHR-FaceSurvival" that can estimate a patient's remaining lifespan based on images of their face.
Scientists have created a tool designed to help doctors gauge the severity of a patient's health condition by examining their face. The American researchers stated that they trained this tool using images of millions of healthy individuals, as well as 35,000 images of cancer patients.
They claim that the tool can estimate how long a patient might live by conducting a detailed analysis of specific areas, particularly the region around the eyes.
According to the researchers, the tool is not yet ready for use in clinical settings; significant further work is required. The scientists explained that the tool is intended to assist doctors in determining the most appropriate treatment for a patient.
Preliminary findings from the study have been published in *The Lancet Health* journal.
Which part of the face provides the most information?
The AI tool is named "FAHR-FaceSurvival." According to the tool, two specific areas of the face are the most critical for estimating an individual's potential lifespan: the area beneath the eyes and the upper part of the nose, located between the eyes.
When estimating a patient's life expectancy, the tool places the greatest emphasis on these specific facial regions. According to researchers, these indicators have proven useful across various types of cancer. The tool has been found effective even in patients with cancer ranging from mild to very severe, including cases where the cancer had spread to the bones or the brain.
Most of the images of cancer patients used to develop this tool came from individuals in the United States and the Netherlands, aged between 50 and 70. However, the study also included some children and patients over the age of 90.
Physicians already conduct various tests and health screenings to assess a patient's current condition and predict their future health status. A patent application has been filed for this AI tool, though it has yet to be approved. Scientists emphasize that the tool is designed to assist physicians rather than replace them.
Dr. Raymond Mak, a lead researcher at Mass General Brigham hospital in Boston, USA, notes that as human life expectancy increases, physicians require additional support to select the most appropriate treatment—whether it be radiation therapy, surgery, or systemic medical treatments.
According to Dr. Mak, the tool helps physicians assess a patient's remaining physical strength and their level of frailty.
Dr. Mak explains that the researchers' goal is to make health assessments based on facial images more accurate, consistent, and applicable to larger populations. This can assist in selecting treatments tailored to each patient's condition, but it does not replace a physician's clinical judgment.
Another AI Tool for Estimating Biological Age
Mac and his colleagues are also testing another AI tool, which estimates an individual's biological age. In other words, rather than determining age based on a date of birth, it assesses how old a person appears based on their physical appearance—specifically their body and face.
This tool is called "FaceAge."
According to the tool, the skin beneath the nose and around the mouth can provide useful indicators regarding the rate at which a person is aging. This includes the lines extending from the nose to the corners of the lips, as well as the lines running from the corners of the mouth toward the chin. Additionally, wrinkles on the forehead and around the eyes, along with the appearance of the temples and cheeks, can offer insights into the aging process.
Dr. Mac notes that the tool has yielded "remarkably consistent" results across various types of photographs taken in everyday settings. Variations in photography techniques, lighting, facial expressions, or the use of makeup did not significantly alter the results.
The tool was also tested in the context of various plastic surgery procedures. In most cases, the results remained largely stable; however, some changes were observed following rhinoplasty—surgery performed to alter the shape of the nose. ‘It’s in the early stages; more data is needed’
Although these results are encouraging, researchers state that more data—specifically images of faces from a diverse range of individuals—is required to accurately evaluate the effectiveness of these tools.
In particular, more images of people from various ethnic groups are needed. Neither tool is fully developed yet; work to further improve them is ongoing.
The research team is also inviting the general public to participate in this study. Interested individuals can send a photograph of their face to the researchers to receive a personal estimate of their biological age.
Professor Miroslav Bober, an expert in video processing and machine learning at the University of Surrey in the UK, notes that it is premature to claim that these tools can monitor the aging process over time or track changes resulting from lifestyle adjustments.
“This is a promising experiment in its early stages, but it is not yet ready for clinical use,” he said.
According to him, the results generated by these tools need to be compared against the professional assessments made by physicians. Furthermore, [data from] various hospitals and... It should also be tested on patients from different groups.
