Summarized by Masters of Longevity from Lifespan.io.
Reports a study using a vision-language AI trained on paired CT images and reports to extract broad physiological and diagnostic signals from routine scans.
Key Takeaways
- The model Percival was trained on over 400,000 CT volumes paired with reports from 52,000 participants.
- Percival showed significant associations with 35 laboratory measurements and 351 diagnosis categories across organ systems.
- Percival outperformed an image-only model for predicting later diagnoses, with strongest results for heart failure, cirrhosis, and acute kidney failure.
