- Researchers at Brigham discovered that cancer patients tend to look approximately five years older than their chronological age and that this advanced facial aging correlates with poorer overall survival outcomes.
Recent advances in artificial intelligence (AI) have ushered in transformative possibilities across the medical field, and one of the latest breakthroughs is the development of FaceAge, an innovative AI tool created by researchers at Mass General Brigham.
The cutting-edge technology predicts cancer survival outcomes with greater accuracy than clinicians by analysing only a patient’s facial photograph.
The implications of FaceAge extend far beyond traditional prognostic methods, offering a non-invasive, objective, and potentially more reliable approach to assessing biological age and aiding clinical decision-making.
At its core, FaceAge estimates an individual’s biological or “FaceAge” based on facial features captured in a simple photograph.
Greater predictive validity
Researchers at Brigham discovered that cancer patients tend to look approximately five years older than their chronological age and that this advanced facial aging correlates with poorer overall survival outcomes.
Notably, FaceAge outperformed clinicians in predicting short-term survival among patients receiving palliative radiotherapy. Whereas clinicians’ survival predictions based solely on patient photos were just marginally better than chance, the AI demonstrated significantly greater predictive validity, underscoring its clinical utility.
The ability of FaceAge to derive meaningful prognostic information from facial images challenges conventional paradigms. Physicians have long recognised that a patient’s appearance may reflect underlying health, yet subjective biases and variability inevitably influence such assessments.
Unlike humans, AI operates without these cognitive biases, enabling an objective analysis that integrates subtle cues imperceptible to the human eye. As co-senior author Dr. Hugo Aerts, director of Artificial Intelligence in Medicine at Brigham, remarked, a simple selfie can harbour clinically critical data that may inform treatment planning and improve patient outcomes.
Furthermore, individuals whose perceived facial age is younger than their actual age tend to experience better responses to cancer therapies, demonstrating the prognostic relevance of this biomarker.
Disease prognosis
The study underpinning FaceAge, published in The Lancet Digital Health, involved training the algorithm on nearly 60,000 images of healthy individuals followed by validation in over 6,000 cancer patient photos from two medical centers.
The rigorous methodology established a robust association between looking older—especially those appearing 85 years or older—and diminished survival, highlighting the potential of facial imaging as a biomarker of biological aging and disease prognosis.
While additional research is necessary to refine the technology and explore its broader applications, FaceAge opens a promising avenue toward biomarker discovery using photographs.
Co-senior author Dr. Ray Mak emphasised that, as chronic diseases are increasingly understood as manifestations of ageing, accurately predicting an individual’s ageing trajectory is paramount.
Moreover, the potential applications of FaceAge extend beyond oncology, envisaging early detection systems across a variety of medical fields, provided these tools are integrated within rigorous ethical and regulatory frameworks.