AI model trained on 3.7 million EHRs predicted cancer a year early
EricTopol · x · 2026-07-27
An AI model trained on routine electronic health records from 3.7 million people reportedly predicted cancer one year before clinical onset.
The study also claims the model improved performance across detection, staging, treatment response, and prognosis. The attached figure shows a unified multimodal transformer, named Oncoformer, that combines longitudinal vitals, lab tests, and imaging to support pan-cancer prediction tasks.
More from Research
- Atlas Discovery exits stealth with a drug-response prediction platform — garrytan · 2026-07-27
- A deep dive traces Kimi K3’s lineage back to GPT-2 across 8 papers and 48 hours — Madisonkanna · 2026-07-27
- METR’s frontier risk report studies misalignment risks inside AI developer orgs — koltregaskes · 2026-07-27
- US Energy Department selects 278 Genesis Mission projects for AI science workflows — thione · 2026-07-27
- Claude Fable 5 reportedly helped disprove an 87-year-old conjecture with a verifiable counterexample — thione · 2026-07-27
- Cisco releases Antares, open-weight models for local code-vulnerability hunting — thione · 2026-07-27