ENIGMA's 300K brain scans reveal scaling laws for AI brain disease diagnosis
PTenigma · x · 2026-09-22
A MICCAI 2026 talk presents the ENIGMA Consortium's pooled dataset of 300,000+ brain scans from 45 countries across 30+ disorders, exploring whether AI can diagnose brain diseases from MRI at 90%+ accuracy and how much data is "enough":
- Simple models win on small samples; VAEs and vision-language models overtake once data is sufficient, with the crossover point predictable via martingale theory and spectral analysis.
- The "Zeta Law of Discoverability" predicts how diagnostic AUC grows with training data based on the data's spectrum.
More from Research
- New Paper Reviews Empirical Evidence on Work and Wellbeing for AI Futures — scychan_brains · 2026-09-22
- Building one of the hardest on-policy lie datasets for Aletheia's Quest lie detection competition — hunarbatra · 2026-09-22
- Question's Gambit lifts deep research agents: GPT-5.5 hits 90.5% on BrowseComp-Plus — omarsar0 · 2026-09-22
- PyroDash cuts inference cost 96% by having a 4B model call the big one only when needed — jiqizhixin · 2026-09-22
- phantom-kv: uncensor LLMs per-request with an 18MB trained KV-cache, no weight edits — Anony6666 · 2026-09-22
- Mathematicians clash over formal proofs: who voted to change math's rules? — jessi_cata · 2026-09-22