Foundation model trained on 40M face photos predicts cancer outcomes and biological age
EricTopol · x · 2026-10-20
Eric Topol highlights a Lancet Digital Health paper on FAHR-Face, a foundation model pretrained with a masked autoencoder on over 40 million face images (WebFace42M), fine-tuned into two task-specific models:
- FAHR-FaceAge: trained on 743,152 age-labeled photos; the gap between facial and chronological age (FAD) serves as a biomarker, validated on independent cancer cohorts (n=38,183 and n=4,892).
- FAHR-FaceSurvival: trained on 34,363 cancer patients and tested on 3,820, using a custom ranking loss to predict survival risk from a single photo.
The study shows face photographs carry AI-detectable health signals linked to cancer outcomes, biological age, lifestyle factors, and survival.
Related event: AI Predicts Cancer Death Risk From a Single Face Photo(2 posts)→
More from Models
- Claimed DeepSeek-V4.1-Flash beats Anthropic on ML research tasks; sandbagging twist — teortaxesTex · 2026-10-20
- DeepSeek beats Claude and Codex on Google Maps scraping speed, at a steep price — teortaxesTex · 2026-10-20
- Bindu Reddy switches from Fable 5.1 to Astra, bets Fable 5.5 will retake the top spot — bindureddy · 2026-10-20
- An AI claimed it ran a calculation it never ran — 'presumption' may explain many AI errors — CourseSome8119 · 2026-10-20
- HeadEdit steers LLM behavior via frozen unembedding matrix, no fine-tuning needed — burny_tech · 2026-10-20
- HealthBench medical AI leaderboard launches: GPT-6 leads, Claude loses 11 points to length penalty — ArtificialAnlys · 2026-10-20