A simple facial photograph may reveal more than appearance. This study shows how tracking subtle changes in facial aging over time could help predict survival and reshape cancer care.
Study: Face aging rate quantifies change in biological age to predict cancer outcomes. Image credit: hedgehog94/Shutterstock.com
A study published in Nature Communications examines the predictive capability of photograph-based face aging rate (FAR) for overall survival in cancer patients.
AI-derived facial age as a measurable biological indicator
Biological aging rates vary substantially between individuals and can influence cancer outcomes independently of chronological age. However, their clinical use remains limited by the lack of practical, noninvasive biomarkers that can be easily applied in routine care.
FaceAge is an artificial intelligence–based tool that estimates biological age from facial features such as skin texture, volume loss, and structural changes. Previous studies have shown that cancer patients predicted to be older than their chronological age have poorer survival outcomes, supporting its potential as a prognostic biomarker.
Using Face Age to measure aging rate
The authors previously developed a model called Foundation Artificial Intelligence Models for Health Recognition (FAHR-FaceAge), which was trained to recognize signs of ill-health on over 40 million facial images. When used with Face Age, they found that patients whose predicted age was five or more years greater than their chronological age had a 21 % higher mortality risk.
Building on this, the researchers examined serial photographs to understand the signs associated with disease progression or treatment response. Such longitudinal measures are already widely used in clinical practice; for example, changes in prostate-specific antigen (PSA) levels over time help assess prostate cancer risk, while variability in blood pressure provides insight into cardiovascular risk.
FAR and overall survival in cancer
The researchers conducted a retrospective study on 2,276 cancer patients on radiation therapy. Most participants were White, with a median age of 63.4 years, and 62.9 % had metastatic cancer at the first radiation therapy course, increasing to 78.7 % at the second.
The researchers used two photographs of each patient, taken as part of routine clinical practice for identification purposes at the start of each radiation therapy course. These were used to predict the biological age using the Face Age artificial intelligence algorithm.
The FAR was calculated as the change in Face Age divided by the time between photographs and provided a measure of the rate of aging. This was analyzed for correlations with overall survival.
The intervals between photographs were categorized into short (10–365 days), midterm (366–730 days), and long (731–1,460 days). The FAR range was very large in the short-term group, because of the small denominator. Thus, only an FAR >20 was reported to be significant in this group, whereas in the mid- and long-term groups, the threshold was set at FAR >10 and >1, respectively.
High FAR is associated with lower overall survival
For many patients, the Face Age predicted a higher than chronological age from the second photograph. A high FAR was associated with poorer overall survival in all groups, after adjusting for time between photographs, sex, race, and cancer diagnosis at the second radiation therapy course.
In the short-term group, the mortality risk was 25 % higher with a high FAR. In the mid-term and long-term groups, a high FAR was associated with a 37 % and 65 % higher mortality risk.
The researchers repeated the analysis with only metastatic cancer patients. The same associations were found, but with more pronounced separation in survival outcomes between groups.
FAR is a stronger predictor of long-term survival outcome
They also examined the combined effects of the initial deviation of predicted facial age from the chronological age (FADRT1) and the FAR. This showed that when both high FADRT1 and FAR were high, the patients invariably had the highest mortality risk.
With increasingly long intervals between photographs, especially in the long-term group, the differences in FAR values become smaller. Even so, FAR becomes the dominant predictor of survival outcomes, even though both measures still play substantial roles in the increased mortality risk.
This indicates that “FAR consistently outperforms FADRT1 as a prognostic marker across all time intervals, with the strongest performance at long-term intervals.”
Possible mechanisms underlying FAR-based prediction
The authors emphasize the nonlinear nature of biological aging, with accelerated molecular aging, such as DNA damage and cellular senescence, often occurring at definite turning points. In cancer patients, such dynamic parameters reflect not only the disease process but also the effect of cancer treatment.
By quantitatively measuring facial aging, FAR could reflect changes in health over the course of treatment. The advantages of using FAR include its accessibility, ease, and cost-effectiveness, allowing repeated measurements to assess changes in health over the course of treatment.
If validated, it could be incorporated into current prognostic parameters to identify patients at high risk across multiple cancer categories, and to guide decision-making regarding monitoring intensity, supportive care, and treatment approaches, particularly in advanced disease settings where less intensive or palliative strategies may be appropriate.
Study limitations
The ethnic/racial and age composition of the sample limits the generalizability of the findings. Moreover, the lack of data on disease progression and treatment meant that the higher FAR could not be interpreted as being causal. Unmeasured factors like cancer cachexia or treatment-related toxicities could have affected the associations observed between FAR and survival.
Because the photographs were taken at specific radiation therapy time points rather than at regular intervals, their use could have introduced indication bias, as different interval groups may reflect distinct clinical scenarios, limiting generalizability. Pending validation of this work, ethical and privacy concerns, and the potential for bias in such facial recognition systems, remain to be addressed before its clinical translation.
Future studies should correlate disease type, stage, and treatment in diverse populations, using easily accessible algorithms with strong data protection barriers in place. The current findings must be validated in prospective studies, and in combination with other markers of aging. If so, FAR could be a tool to help deliver personalized cancer care.
Download your PDF copy by clicking here.
AI model detects pancreatic cancer years before clinical diagnosis