AI Model Analyzes 40 Million Faces to Predict Cancer Mortality Risk
A groundbreaking study reveals that facial analysis can estimate cancer patients' mortality risk, linking perceived age to survival outcomes.
AI trained on over 40 million images links perceived age to cancer prognosis.
Patients appearing older than their actual age face a significantly higher mortality risk.
The study highlights the potential for AI in personalized cancer treatment assessments.
A recent study published in The Lancet Digital Health has unveiled a novel artificial intelligence model capable of predicting mortality risk in cancer patients by analyzing facial features. The research utilized a database of over 40 million photographs to establish a correlation between perceived age and survival outcomes in individuals undergoing cancer treatment.
The study's findings suggest that individuals who appear older than their actual age may face a higher risk of death. Researchers trained their AI system, known as FAHR-FaceAge, on 743,152 portraits with known ages, achieving an average error of 5.1 years in age estimation. When applied to a cohort of 4,892 cancer patients receiving radiotherapy, the average discrepancy increased to nearly seven years, indicating a potential link to biological aging.
In a separate analysis involving 38,183 cancer patients, those who looked at least five years older than their chronological age exhibited a 21% increased risk of mortality. The second AI tool, FAHR-FaceSurvival, was developed using 34,363 photographs and survival data, revealing that patients classified in the highest risk quartile had over three times the mortality rate compared to those in the lowest risk quartile.
The implications of this research extend beyond individual assessments, potentially influencing treatment decisions for older or frail patients. Raymond Mak, a radiation oncologist and lead author of the study, emphasized the goal of enhancing health assessments through quantitative facial analysis, which could support personalized treatment plans without replacing clinical judgment.
Despite the promising results, the study acknowledges limitations, including the need for further validation and the potential for demographic bias. Researchers call for prospective trials to evaluate the AI tools against established prognostic models in clinical settings. This research represents a significant advancement in the intersection of artificial intelligence and healthcare, paving the way for innovative approaches to cancer treatment and patient assessment.
