The latest AI innovation makes it possible to determine biological age from facial photos and achieves astonishing accuracy in predicting cancer prognoses.
Scientists have developed FaceAge, an AI system that can determine a person’s biological age from simple facial photos. The study, published in The Lancet Digital Health, shows that this tool not only accurately determines biological age, but also provides valuable prognoses for cancer patients. The researchers found that, on average, cancer patients have a biological age that is 4.79 to 5 years higher than their chronological age – a difference that correlates with poorer chances of survival.
The system was trained with almost 59,000 facial images of healthy people and tested on over 6,000 cancer patients. The artificial intelligence not only analyzes obvious age characteristics such as wrinkles, but also detects complex patterns such as microvascular changes, fat distribution and structural bone modifications that are related to the systemic state of health.
The clinical validation of the system is particularly noteworthy. In three independent cohorts totaling 6,196 patients, FaceAge showed consistent prognostic power. Each one-decile increase in biological age was associated with a 15.1% higher mortality risk. In a palliative care cohort, the integration of FaceAge into clinical models improved the AUC for predicting 6-month survival from 0.74 to 0.80.
The molecular correlations reinforce the credibility of the system. While chronological age showed no genetic correlations, FaceAge was associated with the expression of the CDK6 gene – a key regulator of cellular senescence. This finding fits with other biomarkers of biological ageing such as epigenetic clocks and telomere length, which are also associated with cancer risk.
Ads
The most important facts about FaceAge
- FaceAge was trained with almost 59,000 facial images and validated on over 6,000 cancer patients
- The system recognizes an average of 4.79 years higher biological age in cancer patients
- In patients with metastatic disease, the age difference increases to 6.2 years
- A biological age of 85 years is associated with a 3.2-fold higher mortality risk
- Integrating FaceAge into clinical decisions reduced the overtreatment rate by 38
- The tool is 20-60× less expensive than alternative biomarkers (about $5 per test)
- FaceAge correlates with molecular markers of cell aging and inflammation
Source: The Lancet
