From PhillyVoice · · 1m 30s
AI predicts cancer survival from single photo
AI model uses face photos to estimate cancer survival chances.
In brief
AI model uses face photos to estimate cancer survival chances. Mass General Brigham's AI model predicts cancer survival from one facial image, with 50% of high-risk patients dying within three months. Originally reported by PhillyVoice.
AI model predicts cancer survival
Researchers at Mass General Brigham developed an AI model that predicts cancer survival from a single facial photo.
Model trained on 40 million photos
The AI foundation model, FAHR-Face, was trained on over 40 million facial images to estimate biological age and cancer survival.
New cancer survival model
FAHR-FaceSurvival predicts cancer survival using 34,000 facial images of patients with cancer.
High-risk patients
Roughly half of cancer patients with the highest FaceSurvival scores died within three months of diagnosis.
Low-risk patients
About 2% of patients with the lowest FaceSurvival scores died within 90 days of diagnosis.
Biological age matters
People with biological age above chronological age had worse cancer outcomes, according to the study.
Physician need
Doctors need better ways to estimate physical reserve before radiation or surgery treatments.
Racial bias concerns
Researchers warn the model could encode racial bias due to historical healthcare disparities.
Expert opinion
Experts say the model's exact mechanisms for predicting survival are not yet fully understood.
Next steps
The model aims to support personalized treatment decisions without replacing physician judgment.
The one thing
Mass General Brigham's AI model predicts cancer survival from one facial image, with 50% of high-risk patients dying within three months.
