AI Predicts Health Arcs: Nature Paper Details 683K Individuals, 52 Years

EricTopol · x · 2026-08-05

Eric Topol details the Nature paper by Sarah Urbut's team. The ALADYNOULLI model integrates EHR and polygenic risk scores, validated on 683,000 individuals across three cohorts, predicting health trajectories for 348 diseases over up to 52 years. It uses Gaussian processes and identifies 20 temporal signatures for disease mechanisms.

Related event: Nature AI Model Predicts Personal Health Trajectories(2 posts)→

Original post →

More from Research

Research channel →