NF AI
Adverse-event extraction from neurofibromatosis clinical-trial literature, harmonized to CTCAE v6 and queryable across studies.
What this is
NF AI automates the slowest step of clinical-evidence review for neurofibromatosis trials β pulling adverse-event tables out of published papers, mapping the free-text terms to a controlled vocabulary, and stacking the result so different drugs and dose regimens can be compared on the same axes.
Today the corpus covers selumetinib, mirdametinib, trametinib, brigatinib, axitinib, and bevacizumab arms across NF1 and NF2. New PubMed candidates land daily through a deterministic poller; eligible papers are extracted, harmonized, and surfaced for human review before joining the database.
How the data flows
What you can do here
Browse Database, Compare, Literature.
Leave comments inline; drop by the design-system page to see the kit.
No signup needed for read; sign in to comment.
Sign in. Once an admin grants the relevant permission you can run extraction, save studies to the database, or approve candidates from the queue.
Default for new accounts is view + comment only β operators grant capabilities per-user from the admin Users panel.
Superuser only.
Admin β Users & permissions to grant per-user capabilities; Monitor for live pipeline state; Settings + Agents for runtime knobs.
References
- Children's Tumor Foundationβ workshop sponsor β CTF AI Workshop, June 27β28 2026.
- NCI CTCAE v6.0β Common Terminology Criteria for Adverse Events; the AE-grade scale this site harmonizes to.
- MedDRAβ System Organ Class hierarchy used for grouping AE terms in Compareβs SOC heatmap.
- Anthropic APIβ Claude vision + text used by the extraction stage; per-candidate cost recorded in the cost_events table.