Scientists use facial images to assess biological aging in cancer treatment

The use of artificial intelligence to analyze external signs of aging is becoming a new tool in precision medicine. Scientists from Mass General Brigham conducted a large-scale study, the results of which were published in the journal Nature Communications. The study was based on the analysis of data from 2,279 patients who underwent radiation therapy between 2012 and 2023. The main goal of the study was to determine how accurately dynamic changes in appearance can serve as an indicator of overall physiological condition and cancer survival prognosis.

The method is based on the use of the FaceAge deep learning algorithm, which estimates a person’s biological age based on a single photograph. Previous research has shown that cancer patients appear five years older than their chronological age on average. The new study expands on this methodology by introducing the Face Aging Rate (FAR) index. This index is calculated based on two or more photographs taken at specific intervals. The researchers divided the change in biological age into time intervals between treatments, allowing for an objective assessment of the rate of wear and tear during the treatment process.

The analysis showed that the rate of facial aging in the examined patients was 40% higher than the rate of chronological aging on average. A stable relationship was established: the higher the FAR value, the lower the probability of long-term survival. The algorithm provided the most accurate predictions when the interval between photographs was two years or more. Notably, the dynamic FAR value proved to be a more reliable prognostic tool than a single measurement of the deviation between biological age and actual age (FaceAge Deviation, FAD). The combination of these two parameters provides a comprehensive picture of a patient’s adaptive capabilities.

From a practical point of view, FaceAge and FAR are cost-effective and non-invasive biomarkers. Integrating these tools into the standard clinical workflow does not require expensive equipment and can be done using standard photographs taken during hospitalization or treatment sessions. This allows doctors to adjust treatment plans in real-time, intensify monitoring for high-risk groups, and provide more accurate patient counseling regarding prognoses.

The research team currently plans to test the system on a wider range of ethnic and age groups. Preliminary tests on a sample of 24,500 patients over the age of 60 have already confirmed that those whose biological age, as estimated by AI, exceeded the actual age by 10 years or more, had significantly worse survival rates. In the future, FaceAge technology can be adapted for monitoring other chronic diseases and assessing the overall health of healthy individuals, as evidenced by the launch of a specialized portal for data collection and further improvement of algorithms.