Every biologic, ranked for the patient in front of you.
64 therapies, 38 evidence-based combinations. 5 guideline societies. 75 PMID-cited sources. The engine resolves them into one transparent, cited recommendation — in under 100 milliseconds.
Therapy selection is getting harder, not easier.
Approved IBD therapies have nearly doubled in three years. Every new comparative study redraws the map — and every patient sits somewhere different on it.
After anti-TNF failure
Risankizumab, ustekinumab, vedolizumab, upadacitinib, guselkumab. Which ranks first here depends on phenotype, comorbidity and prior exposure.
Too many sources to reconcile
ECCO, AGA, ACG, BSG, DGVS — each society, its own wording. Published comparative analyses rarely crown the same winner.
Safety is combinatorial
Elderly with CV risk and latent TB. Pregnant and escalating. The contraindication checklist grows with every new approval.
One engine, all the evidence
Five societies’ guidelines and 75 cited sources, scored into a single ranked list. Not a black box.
Describe the patient. Get a ranked, cited answer.
Every therapy is scored against the specific clinical profile — disease, severity, Montreal classification, comorbidities, prior failures — and returned as a transparent ranking.
Describe the patient
UC or CD, severity, extent (E1–E3) or Montreal (A×L×B), prior failed therapies, comorbidities — TB, pregnancy, CV risk, renal function.
Safety gates run first
Active TB blocks biologics. Pregnancy blocks JAK inhibitors. Heart failure blocks anti-TNF. Before any ranking — no exceptions, no overrides.
Evidence, weighted
Five societies' guidelines and 75 PMID-cited sources are scored for this exact profile — efficacy, safety, onset, durability, route.
A ranked, cited answer
Therapies come back ranked with transparent scores. Every line traces to the guideline paragraph or trial that produced it.
See an example consult.
Seven fixed synthetic cases, each run through the real engine — open one to see its safety-gated, ranked output.
Trusted sources in. Tested output out.
The engine doesn't invent medicine. It reconciles what the field already knows — and proves the output holds up.
Five societies, one weighting
Where guidelines disagree, both positions surface — with the weighting applied, in the open.
Concordance with a clinician-signed oracle
619-item Class-C oracle (617 clinician-signed, 2 proposed; PMID-cited; ECCO/AGA/ACG/BSG/DGVS): 582 assertions executed, 582 passed, 0 failed, 0 vacuous. 528 of them made at least one ranking comparison (Wilson 95% CI 99.3–100%); in 54 rank-order assertions every lower-ranked comparator was absent from the ranking, so they are reported here and not counted in that interval. 37 not executed: 17 microscopic colitis (outside engine scope), 2 proposed and unsigned, 5 retired by clinician ruling (2 whose cited source did not support them, 3 whose defining finding has no engine reader), 13 whose finding has no canonical input carrier (7 upper-GI location L4, 1 two-drug current therapy, 4 infections stated as objects, 1 therapeutic-drug-monitoring block with no reader). Signed as oracle-619-v4 on 2026-10-18: 283 of the executed entries had their input translated to the engine input vocabulary (2026-10-13 and 2026-10-16, assertions unchanged), and the signature covers the translation. Two renal entries (D019, D037) were re-grounded to the EMA SmPC under BUG-2026-1252 and are covered by oracle-619-v4. 4 run under re-grounded assertions. Single signing clinician; independent external evaluation in preparation. Engine 10.1.207, catalog 10.1.31.
Absence assertions (137/137)
All 137 executed 'absent' assertions held: no therapy the oracle says must not appear for that patient, whether a contraindication, a never-event or an agent without a role in that setting, appeared in the ranking. 139 in the fixture, 2 run under re-grounded latent-TB assertions. Wilson 95% CI 97.3–100%.
Built for the people making the call.
Community gastroenterology
A guideline-safe answer inside a 15-minute slot — citation ready for the chart.
IBD centers
Same evidence, same weighting, every clinician. Sequencing decisions you can defend in board review.
Fellows in training
See exactly why #1 outranks #2. The scoring is the curriculum.