Free, attribution-only teaching resources for medical schools, residency programmes and citizen-science courses. Slides, problem sets, case packs and lab exercises — all kept in step with the modules they teach against.
The same AURORA modules sit underneath every track — what changes is the surface, the cases, and the assessment. Pick a track and a module; the pack adapts.
Visual atlas + small-group cases per disease module. Hand-held narrative slides, no clinical reasoning yet — anatomy, mechanism, why the disease exists.
Case packs that surface AURORA's confidence bands explicitly. Students learn to read uncertainty, not just point estimates.
Module-specific decision packs — extent of resection, prenatal repair timing, ETV vs shunt — taught against AURORA's audit trail and override semantics.
Engineering-track pack: how to build on AURORA. PDE solvers, foundation-model heads, replication infrastructure.
Citizen-science-friendly pack: pinned commits, signed weights, replication tracker. The infrastructure of trust.
Co-authored with the patient-advocate council seats. Patient-facing surface is a different artefact, written differently.
Each disease module ships with a teaching pack: slide deck, case set, problem set, and exam-style questions with model answers. Packs ship at public alpha.
Six lab exercises, runnable on a laptop with CPU-only AURORA. Each ships as a Jupyter notebook + answer key, with a small public cohort and an explicit failure-mode test.
Students load a glioma case from the public bench, run GLIO-INFIL, and reason about the uncertainty band on the invasive margin. The lab ends with a misleading case where the band is wide; students discover why.
A patient where MGMT methylation status is ambiguous. Students practice writing a decision packet that surfaces uncertainty instead of hiding it.
Replicate the ETVSS calibration on a public cohort, then compare against AURORA-HYDRO's compartment model. Students see where the priors actually move.
A small cohort of cranio cases with photos and 3D scans. Students train a tiny classifier on top of CRANIO-DIAG and quantify the reduction in CT exposure.
Students intentionally bias a training set and watch the equity loss block release. They then design a stratification scheme that would catch the bias earlier.
Pull a pinned AURORA result with one command. Re-run it. Find the drift. Diagnose whether it's their environment or the AURORA artefact.
Teaching materials are released under CC BY-SA 4.0 at public alpha. Use them in your course, fork them, re-edit them. The only thing the licence asks is that you credit AURORA and share derivatives under the same licence.
If your institution wants to ship a translated, regional or specialty-specific pack on top of these, that's exactly what the licence is for. We'll add your fork to the educator index at public alpha.
We do not teach AURORA. We teach the diseases. AURORA is just the substrate that lets the teaching happen the same way at every hospital.