FaceAge: AI Tool Predicts Cancer Survival from Selfies

Researchers at Mass General Brigham (USA) have developed a AI-driven deep learning tool called FaceAge, that estimates a person’s biological age and can predict survival outcomes in cancer patients using just a facial photograph. This AI-driven system has shown greater accuracy in predicting cancer survival than traditional methods based on chronological age. The breakthrough study was published in The Lancet Digital Health.

Key Highlights

  • Tool Name: FaceAge
  • Developed By: Researchers at Mass General Brigham, USA
  • Purpose:
    • Estimate biological age from a facial photo
    • Predict cancer survival outcomes more accurately than traditional methods
    • Aid personalized cancer treatment planning; all using facial photographs
  • Technology Used: Deep learning, image processing, and machine learning

What is Biological Age?

  • Biological Age reflects the body’s physiological condition. It reflects how a person’s body is aging based on: Overall health status; Genetics; Environment; Lifestyle.
  • Chronological Age is the number of years a person has lived
  • FaceAge focuses on how old a person looks physiologically, not how old they are

How FaceAge Works

FeatureDetails
InputFacial photograph
Technology UsedDeep learning + image processing
Trained on58,851 photos of presumed healthy individuals
Tested on6,196 photos of cancer patients before radiotherapy
PerformanceOutperformed 10 clinicians in predicting survival for palliative cases
Assessment CriteriaSigns of aging (wrinkles, skin tone, etc.) captured in facial features

Key Findings

  • Biological age predicted by FaceAge was often 5 years higher than the actual age in cancer patients.
  • FaceAge outperformed traditional survival prediction methods across 3 patient groups:
    1. Curative – Undergoing potentially curative radiotherapy
    2. Thoracic – With chest cancers (e.g. lung, esophageal)
    3. Palliative – With advanced/metastatic cancer

Why It Matters for Healthcare?

  • Reduces bias in treatment decisions by providing an objective metric of aging.
  • Potential for remote health monitoring and frailty detection in elderly.
  • Helps doctors:
    • Personalize treatment plans
    • Prioritize palliative care
    • Avoid over- or under-treatment
  • Can aid in clinical judgment: structured version of the traditional “eyeball test”.

Potential Applications

AreaUsage
OncologyPersonalizing treatments, predicting outcomes
Clinical TrialsPatient stratification for better data
Remote Health MonitoringNon-invasive, image-based regular health assessments
Geriatric CareAssessing frailty and fall risk in older adults
General HealthEstimating lifespan and health status beyond cancer
Biomarker DiscoveryOpens new ways to discover visual signs of aging and disease

Ethical & Regulatory Concerns

  • Privacy Risks: Facial data misuse by employers, insurers, or authorities
  • Algorithmic Bias: Inaccuracy across races, genders, or age groups if not well-trained
  • Regulation Needed: Oversight to ensure data security, fairness, and transparency

Way Ahead

  • Further research and trials in different hospitals and diverse patient groups
  • Expanded studies on using biological age for wider disease prediction
  • Development of ethical frameworks and data protection laws for AI in healthcare

Connect with our Social Channels

Share With Friends

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top